EXPERIMENTATION OF COMPOSITE REPAIR TECHNIQUES FOR PIPELINES

USING FINITE ELEMENT ANALYSIS WITH RESPECT TO ASME PCC-2, ISO

24817, AND PARAMETRIC MODELS

A Thesis

by

TAYLOR AUSTIN BARRON

Submitted to the Office of Graduate and Professional Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

Chair of Committee, Chii-Der S. Suh

Committee Members, Thomas R. Lalk
Luciana R. Barroso
Head of Department, Andreas A. Polycarpou

August 2016

Major Subject: Mechanical Engineering

Copyright 2016 Taylor Barron

 ABSTRACT

A relatively new method for corrosion repair in steel pipelines has been developed by

externally wrapping damaged pipes with composites. Both ASME PCC-2 and ISO 24817,

engineering standards concerning pipelines and pipe repairs, have analytical solutions to

dictate the minimum composite repair thickness required to safely rehabilitate a corroded

steel pipe. When the pipe is assumed to yield into the composite wrap, certain design

allowances reduce the necessary composite thickness based upon the live pressure of the

composite wrap.

Using finite element analysis, an investigation was carried out to determine whether

ASME PCC-2 and ISO 24817 analytical solutions create code compliant composite wraps

for various steel wall thinning percentages and live pressures. Results indicate that when

considering live pressure, neither ASME nor ISO standards create max hoop strain

compliant composite wraps across all wall thinning percentages and live pressures. A

parametrically modified version of the ASME PCC-2 standard results in code compliant

wraps that more efficiently satisfy the max hoop strain requirement for all tested wall

thinning percentages and live pressures.

It is recommended that ASME PCC-2 equation governing composite repairs on pipes

with live pressure considerations be updated to include parametric modifiers. This will

lead to the creation of more cost efficient, code compliant composite wraps.

ii
 ACKNOWLEDGEMENTS

I would like to thank my committee chair, Dr. Chii-Der Suh, and my committee

members, Dr. Thomas Lalk and Dr. Luciana Barroso for their guidance and support

throughout the course of this research. I would specifically like to thank Dr. Suh for

encouraging me to pursue graduate school. Through this experience, I have gained a

greater appreciation for engineering as a whole and a greater understanding of the research

process.

Secondly, I would like to thank Dr. Christopher Alexander for his guidance in the state

of the art of composite wraps; whose contributions kept this research grounded and

established the initial focus from day one.

I would also like to thank the Civil Engineering Department, Dr. Ray James, Dr. Glen

Miller, and Dr. Martin Peterson for providing an opportunity to continue and enrich my

educational career as a teaching assistant.

Finally, I would like to thank my mother and father for their encouragement and to my

wife for her patience, love, and support.

iii
 CONTRIBUTORS AND FUNDING SOURCES

This work was supported by a thesis committee consisting of Dr. Chii-der Suh of the

Mechanical Engineering Department (advisor), Dr. Thomas Lalk of the Mechanical

Engineering Department, and Dr. Luciana Barroso of the Civil Engineering Department.

This work was funded independently of any primary research resource. Indirect

support of this research was achieved through a teaching assistantship within the Civil

Engineering Department, as well as through a paid summer internship at Stress

Engineering Services.

This thesis is an independent extension of work completed under Dr. Christopher

Alexander while at Stress Engineering Services. Working with Dr. Alexander established

the initial problem statement. All other works were independently decided upon and

completed by the student, including all finite element modelling, analysis, and coding.

iv
 NOMENCLATURE

D Component outside diameter, mm (in.)

c Hoop strain in the composite

 live Hoop strain at live pressure

 clive Hoop strain at live pressure in the composite

 slive Hoop strain at live pressure in the steel

Ec Tensile modulus for the composite laminate in the circumferential direction

determined by test, N/m2 (psi)

Es Tensile modulus for substrate material, N/m2 (psi)

MOWP Maximum Operating Working Pressure SMYS derated as required by the

appropriate construction code) of component, N/m2 (psi)

P Internal design pressure, N/m2 (psi)

Plive Internal pressure within the component during application of the repair, N/m2

(psi)

SMYS Specified Minimum Yield Strength, N/m2 (psi)

s SMYS of the steel component, N/m2 (psi)

ts Minimum remaining wall thickness of the component, mm (in.)

tmin Minimum repair thickness, mm (in.)

v
 TABLE OF CONTENTS

Page

ABSTRACT .......................................................................................................................ii

ACKNOWLEDGEMENTS .............................................................................................. iii

CONTRIBUTORS AND FUNDING SOURCES.............................................................iv

NOMENCLATURE ........................................................................................................... v

TABLE OF CONTENTS .................................................................................................. vi

LIST OF FIGURES ........................................................................................................ viii

LIST OF TABLES ............................................................................................................. x

CHAPTER I INTRODUCTION AND LITERATURE REVIEW ................................... 1

1.1 Overview of Composite Wraps and ASME PCC-2 ................................................. 1

1.2 Literature Review ..................................................................................................... 3
1.1.1 Introduction to Composites in Pipeline Repair.................................................. 3
1.2.2 Installation Pressure Research in Steel Sleeve Repairs ..................................... 5
1.2.3 Analysis of Defect Width in Composite Repairs .............................................. 6
1.2.4 Burst Pressure Modelling in Composite Repair ................................................ 8
1.2.5 Failure Pressure Estimations for Corroded Pipelines ...................................... 10
1.2.6 Live Pressure Analysis of Composite Repairs ................................................ 11
1.2.7 Note on Literature Review Papers Not Presented ........................................... 19
1.2.8 Summation of Literature Review .................................................................... 19
1.3 Review of ASME PCC-2 ....................................................................................... 20
1.3.1 Derivation of ASME PCC-2 Live Pressure Equation ..................................... 20
1.3.2 Indeterminacy of Analytical Solutions ............................................................ 22

CHAPTER II RESEARCH OBJECTIVES..................................................................... 24

2.1 Research Objectives ............................................................................................... 24

2.2 Formulation of Hypothesis ..................................................................................... 25
2.3 Parametric Formulation .......................................................................................... 25

CHAPTER III METHODOLOGY.................................................................................. 28

3.1 Explanation of Variables ........................................................................................ 28

3.2 ASME and ISO Standards Finite Element Model Testing ..................................... 30

vi
 3.3 Finite Element Model Optimization Loop ............................................................. 31
3.4 Parametric Model Matching ................................................................................... 32

CHAPTER IV MODEL DEVELOPMENT .................................................................... 36

4.1 Introduction to the Finite Element Method ............................................................ 36

4.2 Introduction to Finite Element Analysis ................................................................ 37
4.3 Quarter Pipe Analysis............................................................................................. 38
4.3.1 Model Simulation Methodology ..................................................................... 39
4.3.2 Quarter Pipe Model ......................................................................................... 40
4.3.3 Material Properties and Assumptions.............................................................. 42
4.3.4 Forcing Conditions on Quarter Pipe Model .................................................... 44
4.3.5 Boundary Conditions on Quarter Pipe Model ................................................. 44
4.3.6 Elements .......................................................................................................... 45
4.3.7 Convergence Testing ....................................................................................... 46
4.3.8 Model Benchmarking ...................................................................................... 48

CHAPTER V RESULTS ................................................................................................ 50

5.1 ASME and ISO Model Results .............................................................................. 50

5.1.1 Resulting Composite Strain Using ASME PCC-2. ......................................... 53
5.1.2 Resulting Composite Strain Using ASME PCC-2 with No Live Pressure ..... 56
5.1.3 Resulting Composite Strain Using ISO 2481...................................................58
5.1.4 Resulting Composite Strain Using ISO 24817 with No Live Pressure ........... 59
5.2 Results from Optimized Finite Element Model and Best Fit Parametric Model ... 61
5.2.1 Optimized Finite Element Model Results Equaling 0.25% Composite
Strain ........................................................................................................................ 61
5.2.2 Parametric Model Based Off Optimized Finite Element Model Results ........64
5.3 Comparison of Best Fit Parametric Model to ASME PCC-2 and ISO 24817 .......66
5.4 Summary ................................................................................................................ 72

CHAPTER VI CONCLUSIONS AND FUTURE WORK ............................................. 75

6.1 Conclusions ............................................................................................................ 75

6.2 Future Work ........................................................................................................... 76

REFERENCES ................................................................................................................. 77

APPENDIX I EXPLANATION OF THIN WALL PRESSURE VESSELS .................. 80

APPENDIX II FINITE ELEMENT INPUTS FOR EACH TESTED

STANDARDS EQUATION ............................................................................................ 82

vii
 LIST OF FIGURES

Page

Figure 1. Zoomed in section of pipe with filler and composite wrap. ............................... 4

Figure 2. Saeed et al material assumptions. ..................................................................... 14

Figure 3. Visualization of reduced nominal wall thickness ............................................. 29

Figure 4. Comparing thicknesses from optimized finite element model results to

parametric models, various C values. ............................................................... 34

Figure 5. Comparing composite thicknesses from optimized finite element model

results to parametric models, various B values. ............................................... 35

Figure 6. ABAQUS quarter pipe shell model, before loading. ........................................ 40

Figure 7. ABAQUS quarter pipe shell model, fully pressurized. ................................... 41

Figure 8. True stress vs. true strain of X65 steel. ............................................................. 43

Figure 9. Mesh size convergence test. .............................................................................. 47

Figure 10. Comparison of resulting composite thicknesses modelled by ASME

PCC-2 and ISO 24817 equations with and without live pressure. ................... 52

Figure 11. Strain in composite layer modelled with ASME PCC-2 substrate
yielding equation with strain hardening steel. .................................................. 55

Figure 12. Strain in composite layer modelled with ASME PCC-2 substrate
yielding equation with strain hardening and no live pressure. ......................... 57

Figure 13. Strain in composite layer modelled with ISO 24817 substrate yielding
equation with strain hardening.......................................................................... 59

Figure 14. Strain in composite layer modelled with ISO 24817 substrate yielding
equation with strain hardening and no live pressure. ....................................... 60

Figure 15. Strain in composite layer modelled by finite element model

optimization loop. ............................................................................................. 62

Figure 16. Strain in composite layer of the finite element model using the best fit
parametric model, compared to ASME PCC-2 values. .................................... 65

viii
Figure 17. Strain in the composite layer of the finite element model using the best
fit parametric model, compared to ASME PCC-2 no live pressure values. ..... 66

Figure 18. Comparison of composite thickness outputs from best fit parametric
model to ASME PCC-2 equations as well as the optimized finite element
model results. .................................................................................................... 68

Figure 19. Comparison of composite thickness outputs from the best fit parametric
model to ISO 24817 equations as well as the optimized finite element
model results ..................................................................................................... 69

ix
 LIST OF TABLES

Page

Table 1. Material properties of composite laminate and steel pipe .................................. 42

Table 2. Convergence results ........................................................................................... 47

Table 3. Error analysis between analytical solution and ABAQUS model for first
steel step............................................................................................................ 49

Table 4. Resulting ABAQUS inputs from optimization loop that equal 0.25% hoop
strain ................................................................................................................. 63

Table 5. Comparison of strain rates between ASME standard, ISO standard and
parametric model. ............................................................................................. 70

Table 6. Comparison of composite thicknesses between ASME standard, ISO

standard, optimized results, and the parametric model..................................... 71

Table 7. ABAQUS inputs for ASME PCC-2 live pressure considered (0.25% hoop
strain) ................................................................................................................ 82

Table 8. ABAQUS inputs for ASME PCC-2 no live pressure considered (0.25%
hoop strain) ....................................................................................................... 83

Table 9. ABAQUS inputs for ISO 24817 live pressure considered (0.25% hoop
strain) ................................................................................................................ 84

Table 10. ABAQUS inputs for ISO 24817 no live pressure considered (0.25%
hoop strain) ....................................................................................................... 85

x
 CHAPTER I

INTRODUCTION AND LITERATURE REVIEW

The following chapter has three sections. Section 1.1 establishes a general overview

of composite wraps and the variables the testing is interested in. Section 1.2 reviews and

critiques recent literature focusing on pipe repair stresses. Section 1.3 derives and critiques

the ASME PCC-2 equation most central to this research endeavor.

1.1 Overview of Composite Wraps and ASME PCC-2

A relatively new method for corrosion repair in pipelines has been developed by

externally wrapping damaged pipes with composites. Pipes that have external or internal

corrosion where the corroded depth is less than 80% of the steel wall are eligible for

composite over wrap rehabilitation. Both ASME PCC-2 and ISO 24817, engineering

standards concerning pipelines and pipe repairs, have analytical solutions to dictate the

minimum composite repair thickness required to safely rehabilitate a corroded pipe. For

the ASME standard the specific equation in question is PCC-2 Underlying Substrate

Yields Equation 5 (Section 3.4.3.2, 2015) [1]. When the pipe is assumed to yield into the

composite wrap, certain design allowances reduce the necessary composite thickness

based upon the installation pressure of the composite wrap. The following ASME equation

dictates how thick the composite wrap needs to be. The equation is iteratively solved for

the composite thickness, tc, as the hoop strain within the composite,  c , is mandated to be

a maximum of 0.25% hoop strain.

1
 PD t Plive D
c   sy s  (1)
2 Ec tc Ec tc 2( Es ts  Ectc )

In Equation 1 the remaining symbols are final design pressure P, outer steel diameter

Ds, composite Young’s modulus Ec, composite repair thickness tc, yield stress of the steel

Sy, steel thickness ts, live wrapping pressure Plive, and Young’s modulus of the steel Es.

The first of the three terms represents the strain in the composite at the design pressure.

The second term represents the strain held by the thinned (corroded) steel. The third term

represents the strain at the live pressure (composite wrap installation pressure). Two main

variables are examined, live pressure and remaining steel thickness. Live pressure refers

to the installation pressure of the composite upon the corroded steel pipe. Steel thickness

refers to the remaining wall thickness of the steel after the corrosion loss is measured.

Each live pressure/steel thickness scenario creates an input into a finite element model of

the pipe to be examined. More testing detail and the various governing standards are

explained in the methodology section.

2
1.2 Literature Review

The following section outlines the most recent applicable research in composite repair

techniques. Each article is presented, and then critically related to the research methods

used to analyze how live pressure and steel thickness effect the viability of composite

repairs.

1.1.1 Introduction to Composites in Pipeline Repair

Currently, a significant issue facing the long term life of oil and gas transmission

pipelines is managing erosion, corrosion, and unintentional mechanical damage. The

Interstate Natural Gas Association of America [2] estimated that 12% of pipeline

infrastructure was installed prior to 1950, 37% prior to 1960, and 60% before 1970. A

large majority of these pipelines were installed without modern anti corrosion methods

such as interior coatings or cathodic protection. In the United States alone, between 2 and

3.3 billion dollars a year are lost to replacing or repairing corroded pipelines.

Composites offer benefits over traditional welded sleeves because composites do not

require a pipeline to be taken offline. Composite repairs allow for pipeline owners to keep

product moving and reduce opportunity costs losses associated with repairs. Also, there is

no risk of explosion as composite repairs do not create high temperatures like welding

does. Composite repair wraps can be used to rehab steel pipes that suffer from external or

internal corrosion or mechanical damage that does not exceed 80% of the steel’s thickness.

3
 Figure 1 is a cutaway of a composite wrap applied to an in service pipeline. Notice the

pit of corrosion is filled with an epoxy filler before the composite wrap is applied. The

remaining steel thickness is measured from the bottom of the corroded pit.

Figure 1. Zoomed in section of pipe with filler and composite wrap.

Before the installation of a composite wrap, the site is prepped by cleaning the surface

of the pipe to a near white finish and then filling the corroded area with an epoxy filler.

The composites are usually installed in one of two ways. Most repair systems use either

4
carbon fiber or E-glass as the fibers in a resin matrix. There is strong cost savings pressure

to use cheaper E glass over carbon fiber. The fibers are strong (1700 Mpa or 246 ksi) but

brittle, they benefit from being surrounded in a protective matrix that toughens the

resulting lamina. Each fiber is 10 microns in diameter and spun together on the order of

the thousands and woven together into mats. For pipe repair the composite needs to be

primarily strong in the hoop direction (along the circumference). Pipe repair composites

are primarily unidirectional laminates for this reason.

After aligning the fibers, the mat is introduced into a liquid resin that will cure into a

hard matrix around the fibers. When and how the resin is applied usually varies based on

manufacturer. The resin can be partially cured in house and require a final cure in the field

by being exposed to air or water; this is called prepreg. Another way to cure the composite

is to mix the resin on site and apply the resin as the mats are being wrapped around the

pipeline; this is called wet lay up. These methods represent a majority of methods used to

install pipeline composite repairs.

1.2.2 Installation Pressure Research in Steel Sleeve Repairs

Chapetti, Otegui, Mandfredi, and Martin [3] published a full scale analysis of the

stress state in metal sleeve repairs on gas pipelines. While the paper only addresses metal

welding and not composite wraps, it addresses the resulting stresses from applying the

repair to a partially pressurized pipeline. The authors state,

5
 It has been shown that keeping the gas pressure in the pipe near the normal

operating values leads to lower stresses in the sleeves after they are welded…As

soon as the pipe is pressurized again, very high hoop and longitudinal stresses are

added to the sleeves, while the pipe material inside the repair remains almost

unstressed…Although the choice of pressure during repair could be used as a tool

to control stresses in pipe and sleeve and to optimize the integrity of the repair,

pipe pressure during repair welding must be reduced at present due to safety

reasons.

This reasoning is central to the ASME PCC-2 and ISO 24817 standards reduced composite

thickness requirements at higher installation pressures. These findings validate that live

pressure does play a role in the wrap’s stress state. As a whole, Chapetti et al [3] was more

focused on establishing safe steel sleeve procedures like buying high quality materials and

proper welding procedures as this research was commissioned after a catastrophic steel

sleeve failure. Installation pressure findings were just a byproduct of testing, not a main

goal and as such did not present any more findings relating to it.

1.2.3 Analysis of Defect Width in Composite Repairs

Duell, Wilson, and Kessler [4] sought to compare how carbon composite wraps react

to various widths of circumferential defects. They compared finite element models to real

world testing. Their results showed that the width of the defect did not affect the composite

repair’s ultimate burst pressure as the filler putty served as a good strain interface between

6
steel and composite wrap. They also showed that using ANSYS modelling predicted the

burst pressure within 4% of the actual experimental pressure.

While the Duell et al., [4] publication addressed a significant part of pipeline research

that hadn’t been closely examined before, their research methods left something to be

desired. Firstly, their theory section details ASME PCC-2 Substrate Yields equation as

central to dictating the recommended composite thickness for their FEA model and

experimental tests, but then does not use these numbers in either their FEA model or

experimental tests. They use a value that is 25% thinner and likely not to pass code

inspection at design pressure. For the composite thickness they do use (and don’t explain

how they got that number) they laud their FEA model for being within 4% accuracy of

their tests, but do not highlight that they only compared two samples [4]. They give four

sample burst pressures based on the defect width, but only report two experimental results

without mentioning why there are missing data points. Also in their material properties

they explain that their composite laminate material properties test produced +/- 12%

variation with a 95% confidence interval. The small sample size, missing data points, and

large variance in their material properties all detract from their claim that ANSYS came

within 4% of the burst pressure. Furthermore, while their theory section derives ASME

PCC-2 Equation, they do not produce any ANSYS or experimental data to test the

wrapping at live pressure. This paper’s greatest contribution lie within their solid theory

section and ANSYS model. Their application of both leaves more to be desired.

7
1.2.4 Burst Pressure Modelling in Composite Repair

Freire, Veiera, Diniz, and Meniconi [5] published on the effectiveness of composite

repairs with many types of defects as tested with various composite wraps with the goal

of comparing models to real world pressure tests. Freire et al., [5] goals were to understand

how composite layers withstand pressure, compare various reinforcement systems using

experimental testing, and test if modifying the Remaining Strength Factor can be used to

quantify composite wrap quality. Each pressure vessel was subjected to a rigorous

pressurization test simulating what each composite wrap would be expected to perform

across its life. The test cycle is quoted below. Each test vessel was wrapped at its

maximum allowable operating pressure for a defect that was 70% corroded (30%

remaining steel thickness).

The design pressure considers the specified minimum yield strength of the pipe

material decreased to 72%. The second test cycle reached 100% of the pipeline

design pressure and was continually increased to 138.9% thereafter. If the

specimen were able to withstand this pressure for 4 hours, the pressure would be

released to 0 so that external inspection of the specimen could be performed and

that the plastically deformed gages could be reset to zero. The third test cycle

comprised 10 pressure cycles of 100% design pressure. The fourth pressure cycle

consisted of a destructive test for the specimens that survived the first three test

cycles.

8
 The composite wraps of choice ranged from wet hand lay up, to precured, to prepreg

partially precured. No composite wrap passed all the tests for both external and internal

testing. Nonrepaired specimens saw greater burst pressures than would be calculated using

Barlow’s equation and some repairs just marginally beat the non-repaired specimens by

10-20%. The following long quote highlights their results.

Test results showed that three repair systems allowed the pipes to reach the original

design pressure. These repairs withstood pressures above those that ruptured

similar specimens that were not repaired. However, only three of the eleven tested

specimens passed the proposed protocol of pressure tests. Besides, none of the

repair systems were approved in all strength verification tests for both internal and

external defects.

While this conclusion infers that composite wraps can easily withstand burst pressure

tests but not elevated cycling tests, there are some drawbacks to this method. One of the

failure criterion is to pass a four hour 138% maximum operating working pressure

(MOWP) test. Theoretically, all designs should fail a 138% MOWP pressure test as that

is the pressure at which the pipe is designed to burst at. The cycle testing failure analysis

is a valid test though. The 138% MOWP pressure testing showed the substrate steel

yielding and then only elastically deformed when subjected to 100% cyclical tests. This

highlights the elastic theory PCC-2 assumes to be governing the yielded substrate. One of

the biggest concerns about this experiment is the lack of using codes to initially define the

9
composite thickness. For the sake of consistency 25 millimeters was used across all

composites, but the composites had varying material properties, some three times stiffer

in the hoop direction than others. This confounds the failure criterion tests as samples that

were less stiff failed in the most rigorous test the 138% MOWP four hour hold.

Despite the detracting factors, this experiment shines some light on the issue of

ASME’s level of conservative design when considering lifelong failure causes like fatigue

and internal erosion.

1.2.5 Failure Pressure Estimations for Corroded Pipelines

Matherson L. da Silva and Heraldo da Costa Mattos [6] publication they look at the

localized effects of defects when modelling the burst pressure of the pipe. Various defect

sizes and locations were modelled with ASME B31G, RSTRENG .85, and the BG/DNV

failure criterion [6]. These three failure criterion use different assumptions and correction

factors to estimate burst pressure. All of them rely on imprecise, localized measurements

around the defect site. These equations are designed to estimate the remaining strength of

pipes that are not being rehabilitated. The equations were then modified and used to

estimate the burst pressure of pipes that have composite wraps attached to them.

Experiments for both damaged and rehabbed pipes were run.

Their results show that modifying ASME B31G and RSTRENG equations by

including the ultimate tensile stress instead of the flow stress resulted in higher expected

burst pressures than the unmodified equation. ASME B31G tended to most accurately

model the experimental burst pressure for damaged pipes. The modified RSTRENG and

10
the BG/DNV models tended to overestimate the burst pressure, leading to less

conservative designs. For composite wrapped pipes the modified ASME B31G averaged

underestimating the burst pressure by almost a factor of two. The modified RSTRENG

criteria underestimated the burst strength by only 10% [6].

Silva and Mattos [6] published an overall produced a sound report that did an in depth

analysis of localized factors effecting the burst pressure. My only critique of their paper is

not following any code procedure in deciding a composite thickness. As a focused

academic endeavor, the paper did not need to address this issue but its concerns still

remain.

1.2.6 Live Pressure Analysis of Composite Repairs

The final paper to be reviewed is Saeed, Ronagh, & Virk, [7] publication focused on

using finite elements to model how the installation pressure effects the composite

thickness. ASME PCC-2 and ISO 24817 allow for a reduction in thickness of the required

composite as installation pressure increases. The theory is that the composite bears less

load and the underlying steel bears more if the wrap is installed at higher installation

pressures[7]. This is supported directly in the findings of Chapetti et al., [3], and present

in the data in Freire et al., [5]. Saeed et al., [7] was the first publication to examine if the

code’s thickness allowances create composite repairs that are non-compliant with the

codes own guidelines for max hoop strain. Both ASME and ISO dictate that any composite

wrap cannot exceed an averaged hoop strain of 0.25% at design pressure. Using

ABAQUS, Saeed et al., [7] tested a range of live pressures (0%-100%) at incremental,

11
reduced steel wall thicknesses (30-80% missing wall thickness) [7]. Unlike previous

research articles that only looked at individual sections with localized damage and epoxy

fillers, the whole pipe was reduced in steel thickness. This is allowable because the

purpose of this research wasn’t to validate burst pressure calculations, but see if live

pressure had any effects on max strain limitations. Also, previous research by Freire et al.,

[5] support that the stress state of the composite isn’t influenced by the circumferential

width of the defect.

Within Saeed et al., [7] the following equations were used to generate the composite

thicknesses necessary within their finite element model. Each equation is iteratively solved

for the composite thickness, tc, as the hoop strain within the composite,  c , is mandated to

be a maximum of 0.25% hoop strain. As previously explained, reestablished below as

Equation 1 is the ASME PCC-2 Underlying Substrate Yields Equation 5 (Section 3.4.3.2,

2015). Presented as Equation 2 is the corresponding ISO 24817 Underlying Substrate

Yields Equation. Presented as Equation 3 is a derived equation that expands the

circumstances in which ASME PCC-2 Underlying Substrate Yields Equation 6 should be

used to all live pressures.

PD t Plive D
c   sy s  (1)
2 Ec tc Ec tc 2( Es ts  Ectc )

PD t Plive D
c   s y (72%) s 
2 Ectc Ectc 2( Es ts  Ectc ) (2)

12
 PD t
c   sy s (3)
2 Ectc Ectc

In all equations the remaining symbols represent: the final design pressure P, outer steel

diameter Ds, composite Young’s modulus Ec, composite repair thickness tc, yield stress

of the steel Sy, steel thickness ts, live wrapping pressure Plive, and Young’s modulus of

the steel Es. For Equation 2, Sy(72%) refers to 72% of the steel’s yield strength. Two main

variables examined are live pressure and steel thickness. Live pressure refers to the

installation pressure of the composite upon the corroded steel. Steel thickness refers to the

remaining wall thickness of the steel after the corrosion lost is measured.

The ISO equivalent, Equation 2, has the same form as Equation 1 but only allows 72%

of the specified minimum yield stress to be modelled within the steel component. Equation

3 is also mentioned within the ASME and ISO texts as a viable solution for solving for the

necessary minimum composite thickness when the installation pressure is zero. Saeed et

al., (2015) presented a derivation that concluded live pressure plays no part in the

composite strain values, thus Equation 3 was a viable solution for solving for the minimum

composite thickness for any live pressure.

Figure 2 illustrates the material property assumptions for both the steel and composite

on a stress strain diagram. Using elastic perfectly plastic steel and brittle composites,

Barlow’s formula can be used to describe the model. Saeed et al., [7] derivation for

Equation 3 is detailed below.

13
 Figure 2. Saeed et al material assumptions.

The strain in the steel pipe wall at the installation pressure

Plive D
 slive  (4)
2ts Es

The pressure is then increased to an arbitrary value (s1) greater than the live pressure but

less than the steel yielding. The strain in the composite,  c1 , is equal to the strain in the

steel at an arbitrary point before yielding,  s1 , minus the strain in the steel at installation

pressure of the composite wrap,  slive .

 c1   s1   slive (5)

Examining stress equilibrium at point 1, subscript s refers to steel and subscript c

refers to composite:

14
 2( s ts   ctc )  Pint D (6]

)Substituting Young’s modulus and strain for stress in both the composite and steel:

2( Es s1ts  Ec c1tc )  Pint D (7)

Substituting Equation 5 into Equation 7:

Plive D P D
Es ( c1  )ts  Ec c1tc  int (8)
2ts Es 2

Rearranging Equation 8 for strain in the composite:

Plive D P D
Es ( c1  )ts  Ec c1tc  int (9)
2ts Es 2

Pint D Plive D
 c1 ( Ests  Ectc )   (10)
2 2

( Pint  Plive ) D
 c1  (11)
2( Es ts  Ectc )

( Pint  Plive ) D Plive D

 s1   (12)
2( Es ts  Ectc ) 2t s Es

Pressurizing the pipe further to the steel yield point leads to the following equations as P1

and Py are linearly related (before steel yielding).

( Py  Plive ) D
 cy  (13)
2( Es ts  Ectc )

( Py  Plive ) D Plive D (14)

 sy  
2( Es ts  Ectc ) 2t s Es

15
 With the steel and composite relationship defined up to the steel yielding, Saeed et al.,

(2015) focuses on the design pressure. Saeed et al., (2015) defines that after reaching the

steel yield point, any extra pressure is only resisted by the composite. This is due to the

elastic perfectly plastic steel assumption. Furthermore, the steel strains and composite

strains are locked together due to their adhesion. Saeed et al., 2015 continues the derivation

with the following strain assumption. Where  sp equals the strain in the steel at the design

pressure,  sy equals the strain in the steel at steel yielding,  cr equals the strain in the

composite at the design pressure, and  cy equals the strain in the composite at steel

yielding.

 sp   sy   cr   cy (15)

From Barlow’s formula, the design pressure strain can be modeled as such.

2( y ts   c tc )  PD (16)

2( y ts   c Ec tc )  PD (17)

Ultimately, the strain in the composite can be rearranged as such.

PD  y ts
c   (18)
2 Ec tc Ec tc

Which is equivalent to the ASME equation listed previously as Equation 3.

PD t
c   sy s (19)
2 Ectc Ec tc

Saeed et al., [7] model consisted of an axisymmetric 2D model of 3,200 CAX4R

elements (axisymmetric quadrilateral, reduced integration with hourglass controls). The

16
live pressure was modelled by hand calculating the steel’s circumferential expansion due

to the hoop strain at live pressure and then offset the composite wrap by the calculated

amount. The pipe is pressurized from zero psi to the design pressure in one step. Within

the step the steel increases in diameter (due to hoop strain) at the live pressure (via the

offset), makes contact with the composite, and then continues pressurizing onto the design

pressure.

The results from Saeed et al., [7] highlight a few trends. The most noticeable trend is

that using the ASME PCC-2 equation creates non code compliant wraps for all live

pressures greater than 0%. ISO-24817 creates code compliant wraps for most scenarios

but not all while at the same time producing overly thick wraps for increased live wrapping

pressures. The recommended equation (ASME PCC-2 with no live pressure

considerations) creates code compliant wraps, but doesn’t create equally as conservative

wraps across all wrapping pressures or reduced wall thicknesses.

Saeed et al., [7] concludes that his data supports using Equation 3 for all live pressure

scenarios. Furthermore, he concludes that the ASME and ISO live pressure equations are

not conservative enough and need to be revaluated.

After reviewing Saeed et al., [7] paper, two issues arise with their findings. Firstly,

their derivation supported the removal of live pressure considerations by only focusing on

the post steel yielding part of the loading dynamics. Saeed et al., [7] in great detail derives

hoop composite strain up to the yielding point of steel, but does not carry any of these

terms when deriving the post steel yielding behavior. By only examining the post yield

steel behavior it is an obvious solution that live pressure doesn’t play a role, because that

17
section hasn’t been included in the final analysis. In the next section, it is hypothesized

why Saeed et al., [7] analysis only focused on the post steel yielding/design pressure

regime. Secondly, Saeed et al., [7] finite element model (ABAQUS) is vulnerable to

truncation errors. To simulate the composite wrap being joined to the steel surface after

the steel already being pressurized, hand calculations were used to offset the steel from

the composite. Doing any step outside of ABAQUS introduces the possibility of truncation

errors, especially when dealing with small numbers such as .0025 strain rate. Furthermore,

Saeed et al., [7] does not discuss which hand calculations they used. It is assumed they

used Barlow’s thin wall pressure vessel equations instead of Lame’s formula. Lame’s

formula could account for the both the inner and outer wall of the steel and lead to higher

data resolution. If Saeed et al., [7] did use Lame’s formula, it should have been mentioned

in their report. Internally, ABAQUS’s finite element method would produce higher

resolution data more in line with Lame’s formula than Barlow’s formula with a thin wall

assumption.

Ultimately, Saeed et al., [7] presents a truncation vulnerable model and composite load

derivation issues that leave gaps worth investigating. As this is the only published paper

directly analyzing live pressure, it is worthwhile to test Saeed et al., [7] claims

independently.

18
1.2.7 Note on Literature Review Papers Not Presented

There are a few papers important to the school of knowledge on composite wraps not

detailed in this literature review. Their contributions, while important, were not directly

related to research about live pressure installation for composite wraps. Important papers

which discussed the thermal stresses with respect to burst pressure modelling are Esmaeel

et al. [8], Goertzen and Kessler [9], and Mattos, Reis, Paim, da Silva, and Amorim [10].

Creep behavior in carbon fiber/epoxy matrices was examined in detail by Goertzen and

Kessler (2006) [11]. Moisture concerns and effectiveness in water submersion of

composite repairs were addressed by Keller et al., [12], Shamsuddoha, Islam, &

Arayinthan [13], and Alexander and Ochoa [14]. These papers along with the previously

discussed papers represent a well-rounded body of work in composite repairs.

1.2.8 Summation of Literature Review

The papers reviewed outline the current scope of published work focused on

installation pressure of composite wraps for pipelines. Chapetti et al., [3] remarked that

welding repair sleeves at higher internal pressures created less pressure on the repair

sleeve itself. This was highlighted as a positive aspect that could create longer lasting pipe

repairs. Chapetti et al., [3] noted that the high temperatures of welding did not allow for

wrapping at high pressures as that increased the risk of explosion. As composite wraps

became more popular, lots of research publications were produced analyzing the repaired

pipe’s burst pressure. Duell et al., [4] analyzed if circumferential defect width effected

burst pressure, while Freire et al., [5], and Silva and Mattos, [6] independently analyzed
19
various repair techniques, defect sizes, and field applicable predictions of burst pressure.

While Freire et al., [5] mentioned live pressure considerations, it did not play a significant

role in their results and was not the central part of their research. Up until this point,

research tended to focus on better analysis of bursting pressure with no regard to code

conformity. It is evident that researchers time and time again picked a ‘conservative’ or

convenient composite thickness and then back solved for the pressure instead of using the

codes to dictate the composite thickness like field service engineers need to do. Saeed et

al., [7] initially set out to fill this gap in research regarding live wrapping pressure and

hoop strain code compliance, but their research methodology is suspect and could benefit

from more rigorous finite element modelling. This is especially true since Saeed et al.,

[7] calls for a complete overhaul of ASME PCC-2 and ISO 24817 based on their high

strain value calculations.

Given this literature review, there is merit in researching how installation pressure

effect code regulations with more detailed and rigorous modelling methods.

1.3 Review of ASME PCC-2

The following section theorizes how the ASME PCC-2 Equation 1 was derived and

then discusses the limitations of an analytical model.

1.3.1 Derivation of ASME PCC-2 Live Pressure Equation

Equation 1’s approach is to model the composite strain,  c , as a difference in hoop

strain between the final design pressure,  cr , and the live wrapping pressure,  clive . Each

20
of the strain values are taken from hoop stress equilibrium statements. The notation

introduced by Saeed et al., (2015) is used throughout the paper for continuity.

 c   cr   clive (20)

Using a modified version of Barlow’s formula to accommodate the composite and

steel walls, the equilibrium stress equation at the design pressure is:

2( y ts   cr tc )  PD (21)

At the design pressure (ultimate pressure) the stress in the steel has yielded into the

composite and is assumed to be elastic perfectly plastic. The stress in the composite can

be substituted for strain values.

2( y ts  Ec c r tc )  PD (22)

Rearranging for the composite strain at the design pressure reveals:

PD  y ts
 cr   (23)
2 Ec tc Ec tc

Evaluating the equilibrium equation at live pressure:

2( slivets   clivetc )  Plive D (24)

Substituting strain values in:

2( Es slivets  Ec clivetc )  Plive D (25)

For the live wrapping pressure and higher, the steel strain,  slive , and the composite strains,

 clive , are considered equal.

 slive   clive (26)

21
Substituting the composite strain back into the equilibrium equation and rearranging:

2 clive ( Es ts  Ec tc )  Plive D (27)

Plive D
 clive  (28)
2( Es ts  Ec tc )

Plugging Equations 28 and 23 back into Equation 20 results in ASME PCC2 Equation 1:

PD D  y ts Plive D
c    (29)
2 Ec tc Ec tc 2( Es ts  Ec tc )

It should be noted that the derivation of this equation has not been formally released by

ASME. The previous derivation is a belief held by the author about how the committee

arrived at their conclusion. At the very least, there are some reservations about how this

equation was formulated. It does not account for any strain rate variations between the

final design pressure and the live wrapping pressure. Most notably missing is the transition

in strain dynamics as the steel yields.

1.3.2 Indeterminacy of Analytical Solutions

In the literature review section, it was highlighted that Saeed et al., [7] derivation of

composite thickness equation without live pressure considerations only focused on the

design pressure end condition (Equation 19). This was after deriving the composite strain

up to the underlying steel substrate yielding (Equation 13). While Saeed et al., [7] did not

comment on why the composite strain defined up to the steel yield point is not apparent in

the composite strain at design pressure, there is good reason to avoid inserting the pressure

at steel yielding, Py, into any final analytical equation.

22
 ( Py  Plive ) D
 cy  (30)
2( Es ts  Ectc )

The pressure at the steel yield point is unknowable as an input. While the stress and

strain states between the live wrapping pressure and the steel yielding are simple Barlow’s

formulas. The exact pressure at which the steel yields after the composite is applied is

unknowable. It is not an input like Plive, nor is it an output like Pdesign. It exists

indeterminately after the steel and composite have both been stressed at differing levels

depending upon the steel thickness, composite thickness, and live wrapping pressure. The

pressure at which the steel yields is a necessary term to fully model how the composite

and steel dynamically share the load as the steel’s Young’s modulus is no longer a simple

linear model. The change in steel’s ability to withstand loading conditions plays an

undeniable factor in the necessary composite thickness. Given the piecewise nature of the

dynamic conditions, the composite strain at the design pressure could theoretically be

modelled as such.

( Py  Plive ) D ( Pdesign  Py ) D
c   (31)
2( Es ts  Ec tc ) 2( Es , y t s  Ectc )

The Young’s modulus of the steel, post yielding is represented by E s , y . Equation 31 only

serves to prove a point that it is necessary to include the pressure at which steel yields, Py,

as a reference point to account for the change in the composite strain after the steel yields.

This equation still wouldn’t be of practical use since the pressure at steel yield cannot be

independently solved for. There would be two unknowns (composite repair thickness, and

pressure at steel yielding) and one equation.

23
 CHAPTER II

RESEARCH OBJECTIVES

The following chapter consists of three sections. Section 2.1 states the research

objectives, Section 2.2 formulates the hypothesis, and Section 2.3 explains the creation of

a parametric model.

2.1 Research Objectives

Saeed et al., [7] validation effort suffered from two major shortcomings. There are

concerns with relatively large truncation errors because the steel strain at live pressure was

hand calculated and then entered directly into finite element analysis. Furthermore, their

theory stating there is no need for live pressure considerations goes against the body of

literature and brings into question their suspect derivation. Given these issues it is

worthwhile to establish finite element results using a model that simulates all steps inside

the model. Using finite element analysis through ABAQUS, the objective is determine

whether ASME PCC-2 and ISO 24817 result in code compliant composite wraps for

corrosion levels between 30%-80% of the steel wall and live wrapping pressures between

0%-100% of the pipe’s max operating working pressure.

Furthermore, there has been no published effort to establish what composite

thicknesses perfectly relate to 0.25% hoop strain for each live pressure and corrosion

thickness. The results of this research can be used to provide a means to solve for the

composite thickness at each corrosion percentage/live pressure scenario that results in

24
0.25% hoop strain. An additional objective is to create a parametric model that results in

more efficient modelling of composite thicknesses that includes live pressure

considerations.

2.2 Formulation of Hypothesis

Any effort to analytically model the strain in the composite needs to directly address

the model’s dynamic change at the steel yielding point. Since the steel yielding point is

not knowable as an input within an analytical solution, a parametric method is introduced.

Constants were added to each term of Equation 1 (ASME PCC-2 with live pressure

considerations) to best fit the true composite thicknesses that results in 0.25% composite

hoop strain. The true composite thicknesses for each testing scenario that result in perfect

0.25% hoop strain are discerned through running an iterative loop of finite element model

tests with changing composite thicknesses until each steel thickness and live pressure

scenario outputs 0.25% strain.

It is the author’s hypothesis that an parametric adaptation of ASME PCC-2 Live

Pressure (presented as Equation 32) will create composite thicknesses closer to 0.25%

hoop strain at design pressure than ASME PCC-2 or ISO 24817 equations alone.

2.3 Parametric Formulation

The parametric version of Equation 1, listed at Equation 32, introduces constants in

front of every term within the composite thickness equation. These terms allow for each

section’s influence on the governing equation to be visualized. The constants were chosen

25
to be simple values and include higher order terms. Including higher order terms would

introduce inconsistencies and obscure meaning from any findings of better fitting results.

When comparing to the true optimized results, the parametric data was visually inspected

to be better fitting than previous iterations. It was not deemed necessary to run a method

of least squares regression to find the absolute closest match with large amounts of

significant digits. The purpose of this parametric model is to gain general insight into how

the model reacts to various constants. The method of least squares method was also ruled

out because the model should err on the side of overly conservative composite thicknesses.

A simple least squares method could not guarantee that all data points would be slightly

greater than the optimized results, as some points would be lower. Recommending

composite thicknesses thinner than the optimized values would generate non code

compliant wraps that exceeded the 0.25% hoop strain limit. The parametric equation is

presented below.

PD t Plive D
c   *  B * sy s  C * (32)
2 Ectc Ec tc 2( Es ts  Ectc )

Equation 32 dictates how thick the composite wrap needs to be. The equation is iteratively

solved for the composite thickness, tc, as the hoop strain within the composite,  c , is

mandated to be a maximum of 0.25% hoop strain. In Equation 32 the remaining symbols

represent: the final design pressure P, outer steel diameter Ds, composite Young’s

modulus Ec, composite repair thickness tc, yield stress of the steel Sy, steel thickness ts,

live wrapping pressure Plive, and Young’s modulus of the steel Es. The first of three terms

represents the strain in the composite at the design pressure. The second term represents

26
the strain held by the thinned (corroded) steel. The third term represents the strain at the

live pressure (composite wrap installation pressure). Two main variables are examined,

live pressure and steel thickness due to corrosion wall thinning. Live pressure refers to the

installation pressure of the composite upon the corroded steel. Steel thickness refers to the

remaining wall thickness of the steel after the corrosion lost is measured. Each live

pressure/steel thickness scenario creates an input into a finite element model of the pipe

to be examined.

27
 CHAPTER III

METHODOLOGY

The following section is broken up into four sections. Section 3.1 explains the

variables examined within the research. Section 3.2 discusses the differences between the

ASME and ISO standards. Section 3.3 explains the finite element model optimization

process. Section 3.4 illustrates how the parametric model was best fit to the optimized

finite element model.

3.1 Explanation of Variables

The two main variables studied were thickness of steel and live wrapping pressure.

The live wrapping pressure (also referred to as installation pressure) as this represents the

pressure when the composite wrap is installed. The variable representing steel thickness

was declared to be corrosion depth as measured by the % nominal wall thinning (also

referred to as % corrosion). It is assumed within this model that the corrosion could be

modelled as an even removal of steel from the inner layer of steel. Thinning along the

inner wall was selected as to not affect the OD and subsequently Barlow’s stress

calculations. In the real world corrosion can form on the outside of the pipe as well as the

inside, but this model is indifferent to which wall is corroded due to the thin wall pressure

vessel assumption. The finite element model examined various levels of corrosion ranging

from 30% wall thinning to 80% wall thinning. Figure 3 shows how the corrosion was

modelled.

28
 0% Wall Thinning 30% Wall Thinning 80% Wall Thinning

Figure 3. Visualization of reduced nominal wall thickness

The percent wall thinning was adjusted in increments of 10% nominal wall thickness.

For the live wrapping pressure, increments of 25% between 0 MPa and 100% max

operating working pressure (MOWP) were used for each thickness. The Maximum

Operating Working Pressure is 72% of the Specified Minimum Yield Stress of the material

(72% SMYS) For example, X42 steel has a MOWP (72% SMYS) of 1,778 psi at its

nominal thickness of 0.375” and 12.75” outer diameter. When designating wrapping

pressures for each nominal wall thickness, the MOWP is recalculated for each wall

thickness. At 90% nominal wall thickness, the MOWP is 1,600 psi. For 50% nominal wall

thickness the MOWP is 890 psi. For 20% nominal wall thickness the MOWP is 356 psi.

For each nominal wall cases (30%-80% wall thinning), the live wrapping pressure is

designated as 0, 25%, 50%, 75%, and 100% of the steel’s specified MOWP.

29
3.2 ASME and ISO Standards Finite Element Model Testing

Using finite element software package ABAQUS, the ASME PCC-2 and ISO 24817

equations relating to composite thickness with respect to live pressure were examined to

see if their resulting finite element model hoop strains matched the initial inputs of 0.25%

hoop strain. The following four analytical solutions listed below were tested. They

represent the ASME PCC-2 and ISO 24817 live pressure and no live pressure equations.

The only difference between the ISO and ASME standards is that the yield stress in the

steel in ASME standards is the true specified minimum yield stress (SMYS) for the

material while the ISO standard is 72% of the specified minimum yield stress (72%

SMYS). Equations that have been presented before are listed with their original equation

number.

ASME PCC-2 Underlying Substrate Yields, Live Pressure:

PD t Plive D
c   sy s  (1)
2 Ec tc Ec tc 2( Es ts  Ectc )

ISO 24817 Underlying Substrate Yields, Live Pressure:

PD t Plive D
c   s y (72%) s  (2)
2 Ectc Ectc 2( Es ts  Ectc )

ASME PCC-2 Underlying Substrate Yields, No Live Pressure Considerations:

PD t
c   sy s (3)
2 Ectc Ectc

30
 ISO 24817 Underlying Substrate Yields, No Live Pressure Considerations:

PD t
c   s y (72%) s (33)
2 Ec tc Ectc

These four standards equations were each examined over a range of corrosion thicknesses

and live wrapping pressures. For all standards, the steel wall thinning percentages were

[30%, 40%, 50%, 60%, 70%, 80%]. 30% was the minimum as each steel pipe had 1.38

safety factor built in due to the MOWP requirements. Testing below 30% would not

induce steel yielding. 80% wall thinning was the upper limit as that is the maximum

corrosion that can be wrapped using this method, as mandated by ASME and ISO

standards. For each corrosion percentage, five live wrapping pressure percentages were

tested, [0%, 25%, 50%, 75%, 100%]. APPENDIX II displays all the tables illustrating the

inputs for the four tested standards.

3.3 Finite Element Model Optimization Loop

The finite element model in conjunction with python scripting was used to reveal

which composite thickness has exactly 0.25% hoop strain. Input values of steel thickness,

composite thickness, live pressure, and design pressure were inputted into a parameterized

model. The python script outputted the max stress and strain values for the steel and

composite into an output file. The original ASME PCC-2 Equation 1 was used as a

baseline to establish whether the composites needed to be incrementally increased or

decreased in thickness. The optimization loop exited once a live pressure and steel

thickness data point had a composite hoop strain between 0.25% and 0.2505%. It is worth

31
mentioning that the 30% corrosion thickness data points were not optimized as the

necessary composite thickness to establish 0.25% hoop strain ranged between 0.1mm and

0.0001mm. These values were seen as impractical and also unfeasible for pipe repairs.

3.4 Parametric Model Matching

Once the optimized finite element model established the true composite thicknesses,

the parametric version of Equation 1 (presented as Equation 32) was evaluated for best fit.

The following figures illustrate the curve fitting process. Equation 32, the parametrically

modified version of Equation 1, is restated below for convenience.

PD t Plive D
c   *  B * sy s  C * (32)
2 Ectc Ectc 2( Es ts  Ectc )

The C values were adjusted first, with a tested range between 1.50 and 0.05. As the C

value varied, the strain due to live pressure for each given corrosion level varied as well.

Upon inspection, for each corrosion percentage the values only ‘pivoted’ around each 0%

live pressure data point. Figure 4 depicts the C value variations. It was not observed that

all data points translated up or down based on C value modification. At C=1.5, the

subtracted strain due to live pressure tended to overshoot the optimized finite element

model composite thickness, creating unsafe composites. With C=0.05, the subtracted

strain due to live pressure was not enough to match the live pressure reduction rate

produced by the optimized finite element model results. This leads to overly conservative

composite thicknesses. With C=0.20, the subtracted strain due to live pressure tended to

best match the optimized data. It is worth noting for lower corrosion levels (40% and 50%)

32
that no C value properly matched the discounting curve without overestimating the

discount rate at higher wall thinning percentages. This issue is addressed in detail in the

discussion section.

The B values were adjusted between 0.95 and 1.15. Modifying the B values resulted

in a relatively consistent increase or decrease of all strain values. Figure 5 depicts the B

value variations. With B=0.95, all composite thicknesses were overly conservative as

compared to the optimized finite element model results. With B=1.15, all composite

thicknesses were under conservative and produced unsafe composite wraps. By

inspection, B=1.06 resulted in near perfect matching between the parametric model and

the optimized finite element model results for the 0% live pressure data points of each

corrosion level.

The A values were not illustrated, as changing the A value resulted in the same up and

down translations of all data points that modifying the B value created. In an effort to

discern the most knowledge from these parametric evaluations, it was easiest to only

modify A or B and keep the other equal to one.

The best fit parametric model had values of A=1.0, B=1.06, and C=0.20.

33
Figure 4. Comparing thicknesses from optimized finite element model results to parametric models, various C values.

34
Figure 5. Comparing composite thicknesses from optimized finite element model results to parametric models, various
B values.
35
 CHAPTER IV

MODEL DEVELOPMENT

The following chapter discusses the creation of the finite element model using

ABAQUS and details all the steps that went into the model creation. Section 4.1 introduces

the finite element method. Section 4.2 establishes the finite element analysis. Section 4.3

walks through the creation of the quarter pipe with all the necessary assumptions.

4.1 Introduction to the Finite Element Method

The finite element method is used for finding approximate solutions to partial

differential equations through steady state analysis or converting PDE’s into ordinary

differential equations which are solved [15]. The strength of the finite element method

lies in the ability to discretize a complex region (or domain) into a collection of

geometrically simple shapes (elements) [15]. On each element, the governing equation is

formulated using a variational method. The finite element method first systematically

divides the domain into subdomains where each subdomain represents a set of elemental

equations relating to the problem. Once the subdomains are established, an assembly of

elements is created based upon each elements continuity. The global system of equations

is solved using the known general solutions for the problem. Solving each subdomain

results in a matrix with the form [ K e ]{c e }  [ F e ] where each element will have more

unknowns than elemental equations.

36
4.2 Introduction to Finite Element Analysis

Finite element analysis is the consolidation of elements using the finite element

method into solvable matrices with boundary conditions to balance unknown variables.

To evaluate a finite element model, the weak form needs to be established so reduce the

derivative rigor required when solving. To solve for the weak form the approximate

solution is sought over each element. Within a typical finite element  e is assumed to be

n
uhe   u ej ej ( x) (34)
j 1

Where u he are individual solutions of u( x) at the nodes of element  e and  ej

represents the approximation function over the element.

The weak form for a linear analysis can be described as:

xb
 dw du 
B ( w, u )    a
e
 cwu dx (35)
xa  
dx dx

xb

l e ( w)    wfdx  w( x )Q  w( x )Q dx
xa
a 1 b 2 (36)

Where Qn and u represent the boundary conditions at end points a and b.

uhe ( xa )  u1e (37)

 du 
 a   Q1e (38)
 dx  x  xa

uhe ( xb )  u2e (39)

37
  du 
 a   Q2e (40)
 dx  x  xb

In more universal terms, the weak form can be described simply as:

B e ( w, u )  l e ( w) (41)

Solving for the coefficient (or stiffness) matrix for either linear or quadratic elements:

xb
 d ie d ej 
Kije  B e ( ie , ej )    a  c ie ej dx (42)
 dx dx 
xa  

xb

fi e   f
e
i dx (43)
xa

After each element is approximated, a global nodal mesh is created to coordinate local

element nodes to the global nodes. Forces and boundary conditions are introduced and the

K matrix representing the whole model is inverted to solve for the unknowns at each node.

4.3 Quarter Pipe Analysis

The following section outlines the methodology of creating the finite element model

as well as assumptions within the finite element model. The model was required to

replicate the whole simulation of partially pressurizing the pipe, applying the composite

wrap, and then continue internal pressurization to the design pressure. It was not

considered acceptable to have any step done outside of ABAQUS or rely on hand

calculations to simulate a step.

38
4.3.1 Model Simulation Methodology

Simulating the entire loading process entirely within the finite element software

package of ABAQUS was accomplished by using the modelchange() command where the

composite elements were deactivated during the initial loading and then reactivated for

pressurization to the design pressure. The first step takes a bare steel pipe up to the

designated live wrapping pressure. This is performed using the modelchange(), REMOVE

command to remove the composite shell elements. After the pipe is pressurized in the first

step, the composite shell elements are added back as a skin on the steel substrate by using

the modelchange(), ADD STRAIN FREE command. These two steps accurately replicate

the real life procedure of pressurizing the pipe and then adding the composite laminate.

Using this method preserves the resulting stress and strain calculations within ABAQUS.

One run produces all the necessary results while reducing the modeler’s chance of error

by having to enter or subtract strains manually.

The model is created in a python code. This allows access to the modelchange()

command as it is not available in the ABAQUS CAE GUI. The python code is also

parametrized to allow for each iteration to be run based off an input file. The necessary

input variables (steel shell thickness, composite shell thickness, live wrapping pressure,

and ultimate design pressure) are called from a tab delineated line in a text file. After each

39
job is created and ran, the output database is analyzed for the stresses and strains of both

the composite and steel during the live pressure step and the design pressure step. The

python script searches for the element with the max stress for both the composite and the

steel sections. The max steel stress element for the first step is outputted and both the max

stress composite element and max stress steel element for the second step is outputted.

These values are outputted to results text file. The iteration loop continues onto the next

line of input text and repeats the process for all testing needs.

4.3.2 Quarter Pipe Model

The finite element model consisted of a 3D, deformable shell element, 90o revolution

of the steel pipe skinned with a composite laminate. Figure 6 depicts the pipe shell before

loading and Figure 7 illustrates the pipe model after pressurization. Since the diameter to

radius ratio is greater than 10, the thin wall pressure vessel assumption holds and shell

elements were used. A 90 degree shell model was used because axisymmetric models

could not easily accommodate the modelchange() command to simulate the initial steel

only pressurization as well as shared nodes for skins.

40
 Figure 6. ABAQUS quarter pipe shell model, before loading.

Figure 7. ABAQUS quarter pipe shell model, fully pressurized.

41
4.3.3 Material Properties and Assumptions

Table 1 illustrates the material properties for the steel pipe and composite considered for

finite element testing.

Table 1. Material properties of composite laminate and steel pipe

Material Properties of Composite Laminate and Steel Pipe
Composite Laminate Properties
Modulus in hoop direction (E11) 23,800 Mpa
Modulus in axial direction (E22) 24,500 Mpa
Modulus in thickness direction (E33) 11,600 Mpa
Poisson’s ratio (v31) 0.100
Poisson’s ratio (v32) 0.071
Poisson’s ratio (v12) 0.107
Shear modulus (G31) 3,600 Mpa
Shear modulus (G32) 2,600 Mpa
Shear modulus (G12) 4,700 Mpa
Steel Pipe Properties (API 5L X65)
Yield Strength 448 Mpa
Ultimate Strength 530 Mpa
Young’s modulus 200,000 Mpa
Outside Diameter 168.3 mm
Wall Thickness 7.11 mm

The X65 steel is modelled with strain hardening in Figure 8 instead of the standard

elastic perfectly plastic assumption. X65 steel was used as large majority of pipelines use

this steel [16]. This is to gain better insight as to how the yielded steel helps share the load

of the composite repair. The stress strain curve for X65 steel at ambient temperature was

created using ASME Section VIII Div. 2 references.

42
 Figure 8. True stress vs. true strain of X65 steel.

For the composite’s stress strain properties, the material is modelled as perfectly brittle

with catastrophic failure. This is acceptable for uniaxial direction composites used in hoop

stress based repairs. Within the finite element model, the composite itself is modelled as

a homogenous solid with transversely isotropic material properties. Using lamina

properties were not necessary as the composite is not loaded to failure. Both ASME and

ISO standards require composites to have at least 1.0% strain to failure and loading

conditions have a max hoop strain of 0.25%, a fourth of the required minimum strain.

43
4.3.4 Forcing Conditions on Quarter Pipe Model

The pressure was modelled as an internal load applied to interior surface of the quarter

pipe model. In the first loading step, the pipe was pressurized to the live pressure dictated

by the corrosion level and live wrapping percentage. In the second loading step, after the

composite elements have been reactivated, the pipe is pressurized fully to the design

pressure (27.25 MPa).

4.3.5 Boundary Conditions on Quarter Pipe Model

In an effort to make efficient use of the computational resources, only a quarter pipe

was modelled instead of the full circumference of the pipe. Using symmetry boundary

conditions, the finite element model can produce accurate results without the full pipe.

The top and bottom of the pipe are both bound with Y-symmetrical boundary conditions

(U2=0, UR1=0,UR3=0). These two boundary condition makes the effective length of the

pipe three times the original pipe. It is good practice for the length of the pipe modeled to

be at least twice as long as the diameter. These boundary conditions bring the effective

axial length/diameter ratio up to six. The right edge of the pipe (viewer’s orientation on

Figure 6 and 7) is bound by a Z symmetrical boundary condition (U3=0, UR1=0, UR2=0)

this effectively reflects the model another 90 degrees to the right. The left edge of the pipe

(viewers orientation on Figures 6 and 7) is bound by an X symmetry boundary condition

(U1=0, UR2=0, UR3=0). Similarly to the right edge boundary condition, the left edge

boundary condition effectively simulate the whole pipe another 90 degrees to the left.

44
4.3.6 Elements

The quarter section pipe is modeled using a total 4,836 S8R shell elements (standard,

reduced integration, quadratic) for the steel and composite wrap. In order to simulate the

adhesion between the steel and composite, both sets of elements shared a common plane

of nodes. The steel section was assigned to the ‘top plane’ of the shell while the composite

section was assigned to the ‘bottom plane’ of the shell. Shell elements were used over 3D

stress elements as the transverse stress was negligible due to the radius/thickness ratio

being 23.6. Radius to thickness ratios greater than ten are commonly held to satisfy the

thin wall pressure vessel assumption. Plate elements were not used due to the lack of

external bending moments. Standard (implicit) element type was chosen over explicit as

this model is not concerned with the time dependent dynamic steps. The stiffness based

matrix controlling method of standard is better suited for this quasi-static model. The

explicit element solver includes inertial forces and controls the mass matrix with time.

This is not necessary for this model and could lead to unnecessary divergences within the

model. Eight node quadratic elements were selected over four node linear shell elements

to decrease the necessary elements needed to simulate a curved surface. Linear elements

can only draw straight lines between node points and would require and extremely fine

mesh to fully minimize mesh based approximation errors. Quadratic eight node elements

have a node in between each element corner along element edges. This allows for

quadratic approximation methods to fully capture the curved surface of the pipe and

quickly reduce mesh based approximation errors.

45
4.3.7 Convergence Testing

Figure 9 and Table 2 illustrate the mesh convergence test on the shell model. The figure

represents the von Mises stress on the composite shell at the design pressure. The data

point 40% corrosion thickness, 25% live pressure was used. Any data point would have

been suitable as the von Mises values were equivalent across all elements on the shell’s

surface. Both linear shell elements (S4R) and quadratic shell elements (S8R) were

examined to test for convergence due to increasing nodes per element, as well as

increasing elements per model. As expected, S8R elements produced more accurate

results at lower mesh resolutions. This is due to the mid line nodes on the each element

allowing for curved surfaces through quadratic values. Global mesh size of 30, S8R

produced very close results to the fully converged mesh of global mesh size of 5, S8R.

Though, (5,S8R) was selected to perform the testing analysis. (2.5, S8R) was extremely

resource taxing and did not improve the outputting data over the less taxing (5, S8R) mesh

and element combination.

46
 Mesh Size Convergence Test
52.8

52.6

52.4

52.2

52

51.8

51.6

51.4
20,(S4R) 10,(S4R) 5, (S4R) 40,(S8R) 30,(S8R) 20,(S8R) 15,(S8R) 10,(S8R) 5,(S8R) 2.5,(S8R)

Figure 9. Mesh size convergence test.

Table 2. Convergence results

Convergence Results
Von Mises
Mesh size, (element)
Stress (Mpa)

20,(S4R) 51.9162
10,(S4R) 52.593
5, (S4R) 52.593
40,(S8R) 52.633
30,(S8R) 52.6434
20,(S8R) 52.6478
15,(S8R) 52.6481
10,(S8R) 52.6415
5,(S8R) 52.6416
2.5,(S8R) 52.6416

47
4.3.8 Model Benchmarking

Currently, only one published paper directly examines this research problem, Saeed et

al., [7]. The published paper was not used as a definitive benchmark test due to research

group’s suspect finite element model. Furthermore, real world data that can be used to

benchmark this model poses multiple issues. Firstly, this specific area of research is

relatively new and live testing of composite wraps is reviewing papers from a small sample

size. Of those composite wrap real world experiments, most research groups tended to

focus on testing burst pressure models instead of live wrapping pressures. The remaining

groups focusing on live pressure scenarios exclusively tested by machining localized

defects instead of testing consistently thinner pipe like this model assumes.

This leaves benchmarking a computer generated model by analytical solutions. As

already established in Chapter I, this specific model cannot be accurately analyzed using

analytical solutions due to the steel yielding term being a fundamental lynchpin for

describing strains.

The resolve the issue of error analysis, the hoop stress in the steel-only first step of the

model was compared to the exact analytical solution (Barlow’s formula) to see the error

between the analytical stress and the elemental stress. As illustrated in Table 3, the finite

element model averaged 0.00057% difference across all sample inputs tested when

compared to the max stress elements to the Barlow’s formula solutions. Barlow’s formula

is listed as Appendix I for reference.

48
Table 3. Error analysis between analytical solution and ABAQUS model for first steel step
Error analysis between analytical solution and ABAQUS mode for
first steel step
ABAQUS
Analytical
Wall Live Model
Hoop Percent
Thinning Pressure Hoop
Stress Error (%)
(%) (%) Stress
(Mpa)
(Mpa)
0% 0 0 N/A
25% 80.62980 80.62956 0.000305%
30% 50% 161.25791 161.25742 0.000305%
75% 241.88771 241.88696 0.000311%
100% 322.51582 322.51483 0.000308%
0% 0 0 N/A
25% 80.62896 80.62869 0.000336%
40% 50% 161.25791 161.25737 0.000336%
75% 241.88687 241.88606 0.000334%
100% 322.51582 322.51474 0.000336%
0% 0 0 N/A
25% 80.64000 80.63955 0.000559%
50% 50% 161.28000 161.27933 0.000415%
75% 241.92000 241.91792 0.000860%
100% 322.56000 322.55725 0.000853%
0% 0 0 N/A
25% 81.36867 81.36832 0.000426%
60% 50% 161.25791 161.25723 0.000423%
75% 241.88687 241.88585 0.000420%
100% 322.51582 322.51447 0.000419%
0% 0 0 N/A
25% 80.64000 80.63920 0.000995%
70% 50% 161.28000 161.27879 0.000750%
75% 241.92000 241.91838 0.000670%
100% 322.56000 322.55795 0.000636%
0% 0.0 0.0 N/A
25% 80.64000 80.63905 0.001174%
80% 50% 161.28000 161.27870 0.000806%
75% 241.92000 241.91776 0.000926%
100% 322.56000 322.55740 0.000806%
0.000571% Average

49
 CHAPTER V

RESULTS

The following section presents and discusses the results from the all finite element

models tested. There are three sections, Section 5.1 discusses the ASME PCC-2 and ISO

24817 testing. Section 5.2 discusses the optimized finite element model results and the

best fit parametric model results. Section 5.3 compares the best fit parametric model

results directly to the ASME PCC-2 models and the ISO 24817 models.

It is worth noting that 30% wall thinning results, while considerable in the real world,

did not create good data for comparing models. Pipes are generally designed to operate at

a max 72% specified minimum yield stress (SMYS) for safety reasons. Simulating 30%

steel thinning produces uncharacteristically low results as compared to the rest of the data

points as the steel has barely yielded if at all. For this research endeavor 30% corrosion

results are shown for presentation’s sake when applicable but not included in the

ABAQUS optimization loop efforts.

5.1 ASME and ISO Model Results

Both the ASME and ISO standards provide equations for solving for the necessary

composite thickness with live pressure considerations as well as without. These four

solutions all suggest different composite thicknesses for the same loading conditions.

Figure 10 shows how varied each of the four solutions are across various corrosion

thicknesses and installation pressures. ISO 24817, No Live Pressure Considered produces

50
the most conservative composite out of the four solutions for all testing scenarios. The

ASME PCC-2 with Live Pressure solution produces the least conservative composite wrap

out of the four solutions. At relatively little corrosion (30%), the ISO equations are

extremely conservative compared to the ASME PCC-2 equations. This is a result of the

ISO equations artificially reducing the max stress carried by the steel from the actual yield

stress to 72% yield stress. Modifying the max stress the steel can handle only serves to

make the ISO formula a safety factor based design instead of attempting to match the

originally inputted hoop strain. For the ASME standard, as the steel becomes thinner, the

level of live pressure discounting of the composite thickness increases. At 40% corrosion,

the subtracted strain due to the live pressure is considerably less than at 80%. Table 4 at

the end of this chapter documents the exact strain values of all the standards and the

parametric model.

51
Figure 10. Comparison of resulting composite thicknesses modelled by ASME PCC-2 and ISO 24817 equations with
and without live pressure.
52
5.1.1 Resulting Composite Strain Using ASME PCC-2.

In Figures 11 through 14, the resulting strains from the ASME and ISO composite

thickness minimums illustrated in Figure 16 are presented. Each hoop strain in Figures 1

through 14 are outputs of finite element models for each corrosion and live pressure

scenario as the pipe is pressurized fully to the original design pressure. If any of the ASME

and ISO solutions were perfect, the finite element results would show consistent 0.25%

hoop strain for every corrosion and live pressure scenario. This is not the case for any of

the four models (ASME and ISO with and without live pressure). Only a few data points

across the four models actually match the 0.25% hoop strain input for the governing

equation. The following sections highlight how each model over estimates or

underestimates 0.25% hoop strain. Models that show strains below 0.25% indicate that the

composite for that specific corrosion level and wrapping pressure was overly thick. Strains

above 0.25% hoop strain indicate that the composite thickness was too thin. Too thin

composite thicknesses produces non code compliant composite wraps and increases the

risk of pipe rupture. Too thick composite thicknesses cut into the economic advantages of

composite wraps.

For Figure 11, the ASME PCC-2 Substrate Yielding equation (Equation 1) has less

than 0.25% hoop strain at zero live pressure for each steel corrosion level. This means, for

those data points, the composite thicknesses were overly conservative and too thick.

Regarding the live pressure considerations, for 30% and 40% wall thinning wrapping at

any live pressure creates increasingly over conservative thicknesses. This means that the

53
ASME PCC Substrate Yielding Equation does not discount the composite thickness nearly

enough as the wrapping pressure increases. It is worth noting that for 30% wall thinning

the recommended composite thickness is between 2.0 and 0.4 millimeters. The 30% wall

thinning results can be disregarded as explained in the beginning of this chapter. For 50%-

80% steel wall thinning there is a reversal on the effects of live pressure. The transition

from creating more conservative wraps to creating less conservative wraps happens

between 40% and 50% steel wall thinning due to corrosion. For 50%-80% wall thinning,

an increase in wrapping pressure corresponds with a decrease in the conservativeness of

the pipe. Some specific live wrapping pressures result in near perfect 0.25% strain

matching (60% wall thinning, 50% live wrapping pressure), but this is quickly

overshadowed as a significant amount of test cases exceed the 0.25% hoop strain

maximum. The ASME PCC-2 Equation with live pressure considerations should not be

recommend for pipe repairs as it is not conservative enough during high corrosion wraps

and is also too conservative for lesser corroded pipes.

54
Figure 11. Strain in composite layer modelled with ASME PCC-2 substrate yielding
equation with strain hardening steel.

55
5.1.2 Resulting Composite Strain Using ASME PCC-2 with No Live Pressure

Figure 12 illustrates the resulting hoop strains in the finite element models created with

the ASME PCC-2 solution with no live pressure considerations. The results include a

shadow of the Figure 11 results that include the live pressure considerations. This is so

comparisons between the PCC-2 models with and without live pressure can be made.

Without discounting for live pressure, all wall thinning percentages see a drop in hoop

strain as the wrapping pressure increases. All tested scenarios are less than 0.25% hoop

strain, but there is room for improvement. Testing without live pressure considerations

highlights the fact that installation pressure plays a role in developing the necessary

composite thickness. If installation pressure didn’t, the composite hoop strain would be

equal across all live pressures for a given corrosion level. The results show that wrapping

at higher wrapping pressure results in a need for a thinner composite as the hoop strain

decreases. Between no live pressure consideration and PCC-2’s recommendation lies a

composite thickness that can closely match 0.25% hoop strain.

56
Figure 12. Strain in composite layer modelled with ASME PCC-2 substrate yielding
equation with strain hardening and no live pressure.

57
5.1.3 Resulting Composite Strain Using ISO 24817

Figure 13 illustrates the resulting hoop strains in the finite element model created with

the ISO 24817 solution with live pressure considerations. The results include a shadow of

the Figure 11 results that include the live pressure considerations. This is so comparisons

between the PCC-2 and ISO 24817 models can be made. The ISO solution follows the

same trends as the PCC-2 live pressure equation. Both show transitions of

conservativeness between 40% and 50% wall thinning. The ISO standard produces results

under 0.25% hoop strain for every testing scenario except the most extreme case (80%

wall thinning, 100% live wrapping pressure). While not all tested scenarios produce less

than 0.25% hoop strain, all scenarios are less than the ASME PCC-2 counterpart. This is

due to the only difference between the ISO and ASME standards; the ISO standard limits

the max stress allowed to be carried by the steel. This difference creates thicker composites

and thus less strain within each composite compared to each scenario’s PCC-2

counterpart.

58
 Figure 13. Strain in composite layer modelled with ISO 24817 substrate yielding
equation with strain hardening.

5.1.4 Resulting Composite Strain Using ISO 24817 with No Live Pressure

Figure 14 illustrates the resulting hoop strains in the finite element model created with

the ISO 24817 solution without live pressure considerations. The results include a shadow

of the Figure 11 results that include the live pressure considerations. This is so

comparisons between the PCC-2 live pressure and ISO 24817 no live pressure models can

be made.

For every wall thinning and live pressure scenario tested, the ISO no live pressure

model resulted in the most conservative composite wraps with lesser composite strain than

the ASME live pressure model. While this is the only model meeting the minimum

requirements of creating code compliant composite wraps, the ISO no live pressure model

59
is overly conservative for all scenarios. Furthermore, as the wrapping pressure increases,

each resulting composite thickness becomes increasingly overly conservative.

Figure 14. Strain in composite layer modelled with ISO 24817 substrate yielding
equation with strain hardening and no live pressure.

60
5.2 Results from Optimized Finite Element Model and Best Fit Parametric Model

All four of the standards solutions did not produce 0.25% hoop strain values for all

wall thinning levels and live wrapping pressures. The finite element software package

ABAQUS and python scripting were used to search for the necessary composite thickness

to equal 0.25% hoop strain. The resulting composite thicknesses were then used to fit a

parametric model with constants in front each part of the ASME PCC-2 Live Pressure

equation (Equation 32). Each of the constant’s values were changed to output thickness

values that best fit the optimized finite element model results that had perfect 0.25% hoop

strain.

It was found that the parametric version of ASME PCC-2 Live Pressure created the

best fit with values of A=1, B=1.06, C=0.20. The following section presents the results

from the optimized finite element model and the best fit parametric model. The parametric

equation (Equation 32) is reestablished below for easier referencing.

PD t Plive D
c   *  B * sy s  C * (32)
2 Ectc Ec tc 2( Es ts  Ectc )

5.2.1 Optimized Finite Element Model Results Equaling 0.25% Composite Strain

Figure 15 illustrates the resulting hoop strains in the finite element model created with

optimized composite thickness loop. The results include a shadow of the Figure 11 results.

Optimized results for 30% wall thinning were not presented as the finite element model

61
optimization loop was recommending composite thicknesses less than 0.1 millimeters.

30% corrosion does not represent enough steel thinning to warrant a composite wrap

solution using these equations. Table 4 showcases the finite element optimization loop

results that generated perfect 0.25% hoop strains across every live pressure and wall

thinning scenario.

Figure 15. Strain in composite layer modelled by finite element model optimization
loop.

62
Table 4. Resulting ABAQUS inputs from optimization loop that equal 0.25% hoop strain
Resulting ABAQUS inputs from optimization loop that equal 0.25% hoop strain
Live
Corrosion Steel Composite Live Design
Pressure
Percentage Thickness Thickness Pressure Pressure
Percentage
(%) (mm) (mm) (MPa) (MPa)
(%)
0% 4.977 - - -
25% 4.977 - - -
30% 50% 4.977 - - -
75% 4.977 - - -
100% 4.977 - - -

0% 4.266 4.3188 0.0 27.25

25% 4.266 3.3337 4.0875 27.25
40% 50% 4.266 2.6519 8.1750 27.25
75% 4.266 2.1589 12.2625 27.25
100% 4.266 1.7642 16.3500 27.25

0% 3.555 9.9200 0.0 27.25

25% 3.555 9.1000 3.4067 27.25
50% 50% 3.555 8.5158 6.8134 27.25
75% 3.555 8.1400 10.2201 27.25
100% 3.555 7.8200 13.6269 27.25

0% 2.844 15.5900 0.0 27.25

25% 2.844 14.9034 2.7500 27.25
60% 50% 2.844 14.4469 5.4500 27.25
75% 2.844 14.1000 8.1750 27.25
100% 2.844 13.9000 10.9000 27.25

0% 2.133 21.2000 0.0 27.25

25% 2.133 20.6500 2.0440 27.25
70% 50% 2.133 20.2692 4.0881 27.25
75% 2.133 20.1199 6.1321 27.25
100% 2.133 19.8728 8.1761 27.25

0% 1.422 26.8325 0.0 27.25

25% 1.422 26.4900 1.3627 27.25
80% 50% 1.422 26.2176 2.7254 27.25
75% 1.422 26.1000 4.0881 27.25
100% 1.422 25.9000 5.4507 27.25

63
5.2.2 Parametric Model Based Off Optimized Finite Element Model Results

Figures 16 and 17 illustrate the resulting hoop strains in the finite element model

created from the best fit parametrically modified ASME PCC-2 solution with live pressure

considerations (Equation 32, where A=1, B=1.06, C=0.2) . The results include a shadow

of the Figure 11 results that include the live pressure considerations. Figure 16 is presented

against the ASME PCC-2 Live Pressure values, while Figure 17 is presented against the

ASME PCC-2 No Live Pressure values. In Figure 16, the best fit parametric model does

a better job of decreasing composite thicknesses that are under 0.25% hoop strain, and

increasing the composite thicknesses for values over 0.25% hoop strain than any of the

original standards analytical solutions. It is worth noting that the parametric formula does

not perfectly match the optimized finite element results and thus does not result in perfect

0.25% hoop strains across all corrosion and live pressure levels. The 40% wall thinning

scenario has the furthest departure from 0.25% hoop stress. This is expected as the 30%

and 40% wall thinning sections are the steel thicknesses most strongly influenced by the

steel yielding. The 30% and 40% wall thinning sections have the largest percentage of

composite straining between the live wrapping pressure and the steel yielding. This is due

to the excess of remaining steel material that prevents steel yielding at lower pressures

when compared to thinner wall thicknesses. As stated previously, the inability to know the

pressure at steel yielding directly will hamper analytical models from fully describing the

composite’s hoop strain at all wall thinning percentages.

64
 Figure 17 shows a direct comparison of the best fit parametric model’s improvements

over the ASME PCC-2 No Live Pressure solution. For all comparisons within Figure 23,

the parametric model offers a composite wrap more closely aligned with 0.25% hoop

strain.

Figure 16. Strain in composite layer of the finite element model using the best fit
parametric model, compared to ASME PCC-2 values.

65
Figure 17. Strain in the composite layer of the finite element model using the best
fit parametric model, compared to ASME PCC-2 no live pressure values.

5.3 Comparison of Best Fit Parametric Model to ASME PCC-2 and ISO 24817

As a visual, Figures 18 and 19 compare the output thickness of the best fit parametric

model to the optimized finite element results and both ASME and ISO standards. The best

fit parametric model tracks closer to .25% hoop strain more so than any of the ASME or

ISO standards equation. Table’s 5 and 6 showcase all the tested theories hoop strains and

composite thicknesses side by side. Equation 32 resulted in all of the testing scenarios

safely conforming to the max hoop strain limit. Furthermore the best fit parametric model

produced the closest values to 0.25% hoop strain across all wall thinning percentages and

66
live pressures. There is still not perfect agreeance between the best fit parametric model

and consistent 0.25% hoop strains across every walling thinning level and wrapping

pressure. The parametric model still produces considerably overly conservative composite

wraps for greater than 50% live wrapping pressures at 40% wall thinning. Though, the

parametric model was still the best performing method for 40% wall thinning, it still did

not fully replicate the nature of the loading dynamics. The lack of hoop strain matching at

30% and 40% wall thinning rates points back to the model’s inability to establish a

pressure at steel yielding term. It is theorized that the 30% and 40% wall thinning

percentages have considerable amounts of composite straining before the steel yields. All

of the tested models do not establish clear methods of accounting for composite strain

after live pressure but before steel yielding. The best fit parametric model had B values of

1.06 and C values of 0.20. These rates tend to suggest that the ASME codes need to

reexamine the standards to account for these inconsistencies. It is theorized that the steel

yielding term, Ys, should be increased to a steel stress equivalent to no greater than 0.25%

strain. This would better reflect the 6% increase in the steel stress term the parametric

model recommends. As to why the C values need to be reduced to 20% of the current

term, there is not as obvious of an answer at this time. Given no straightforward answer,

the author chooses not to speculate but consider it a starting point for further research.

67
Figure 18. Comparison of composite thickness outputs from best fit parametric model to ASME PCC-2 equations as
well as the optimized finite element model results.

68
Figure 19. Comparison of composite thickness outputs from the best fit parametric model to ISO 24817 equations as
well as the optimized finite element model results

69
Table 5. Comparison of strain rates between ASME standard, ISO standard and parametric model.
Comparison of strain rates between ASME standard, ISO standard and parametric model
Empirical
ASME PCC- ASME PCC- ISO 24817, ISO 24817,
Live PCC-2, Live
Corrosion 2, Live 2, No Live Live No Live
Pressure Pressure
Percentage Pressure Pressure Pressure Pressure
Percentage (A=1, B=1.06,
(%) strain strain strain strain
(%) C=.20) strain
(mm/mm) (mm/mm) (mm/mm) (mm/mm)
(mm/mm)
0% 0.2138% 0.2138% 0.1680% 0.1680% -
25% 0.1798% 0.1786% 0.1420% 0.1387% -
30% 50% 0.1451% 0.1434% 0.1146% 0.1094% -
75% 0.1102% 0.1084% 0.0864% 0.0806% -
100% 0.0749% 0.0734% 0.0580% 0.0525% -

0% 0.2214% 0.2214% 0.1761% 0.1761% 0.2473%

25% 0.2055% 0.1962% 0.1550% 0.1500% 0.2257%
40% 50% 0.1921% 0.1732% 0.1344% 0.1249% 0.2057%
75% 0.1803% 0.1526% 0.1174% 0.1021% 0.1872%
100% 0.1694% 0.1339% 0.1039% 0.0834% 0.1702%

0% 0.2282% 0.2282% 0.1852% 0.1852% 0.2472%

25% 0.2278% 0.2121% 0.1709% 0.1634% 0.2364%
50% 50% 0.2326% 0.1994% 0.1632% 0.1453% 0.2285%
75% 0.2408% 0.1889% 0.1615% 0.1321% 0.2228%
100% 0.2509% 0.1805% 0.1637% 0.1225% 0.2185%

0% 0.2339% 0.2339% 0.1955% 0.1955% 0.2471%

25% 0.2410% 0.2240% 0.1906% 0.1803% 0.2419%
60% 50% 0.2532% 0.2169% 0.1937% 0.1702% 0.2392%
75% 0.2682% 0.2113% 0.2012% 0.1633% 0.2378%
100% 0.2854% 0.2070% 0.2109% 0.1585% 0.2378%

0% 0.2384% 0.2384% 0.2077% 0.2077% 0.2471%

25% 0.2476% 0.2326% 0.2102% 0.1992% 0.2448%
70% 50% 0.2605% 0.2285% 0.2180% 0.1940% 0.2442%
75% 0.2757% 0.2255% 0.2283% 0.1905% 0.2447%
100% 0.2926% 0.2231% 0.2404% 0.1879% 0.2458%

0% 0.2419% 0.2419% 0.2213% 0.2213% 0.2471%

25% 0.2499% 0.2388% 0.2265% 0.2172% 0.2463%
80% 50% 0.2600% 0.2367% 0.2343% 0.2148% 0.2465%
75% 0.2714% 0.2350% 0.2435% 0.2130% 0.2474%
100% 0.2841% 0.2338% 0.2538% 0.2117% 0.2486%

70
Table 6. Comparison of composite thicknesses between ASME standard, ISO standard, optimized results, and the parametric model
Comparison of composite thicknesses between ASME standard, ISO standard, optimized
results, and the parametric model
Empirical
ASME ASME ISO ISO
PCC-2, Live
Live PCC-2, PCC-2, No 24817, 24817, No
Corrosion ABAQUS Pressure
Pressure Live Live Live Live
Percentage Optimization (A=1,
Percentage Pressure Pressure Pressure Pressure
(%) Results B=1.06,
(%) strain strain strain strain
C=.20) strain
(mm/mm) (mm/mm) (mm/mm) (mm/mm)
(mm/mm)
0% 1.0654 1.0654 11.5581 11.5581 - -
25% 0.9202 1.0654 10.2324 11.5581 - -
30% 50% 0.8093 1.0654 9.1391 11.5581 - -
75% 0.7220 1.0654 8.2310 11.5581 - -
100% 0.6516 1.0654 7.4699 11.5581 - -

0% 6.4188 6.4188 15.4125 15.4125 4.3188 4.4916

25% 5.6337 6.4188 13.8053 15.4125 3.3337 4.3661
40% 50% 5.0029 6.4188 12.4349 15.4125 2.6519 4.2467
75% 4.4889 6.4188 11.2657 15.4125 2.1589 4.1330
100% 4.0642 6.4188 10.2651 15.4125 1.7642 4.0248

0% 11.7722 11.7722 19.2670 19.2670 9.9200 10.1662

25% 10.5178 11.7722 17.4882 19.2670 9.1000 9.9259
50% 50% 9.4558 11.7722 15.9178 19.2670 8.5158 9.6941
75% 8.5549 11.7722 14.5365 19.2670 8.1400 9.4705
100% 7.7877 11.7722 13.3236 19.2670 7.8200 9.2547

0% 17.1256 17.1256 23.1215 23.1215 15.5900 15.8408

25% 15.6034 17.1256 21.3051 23.1215 14.9034 15.5371
60% 50% 14.2469 17.1256 19.6441 23.1215 14.4469 15.2405
75% 13.0435 17.1256 18.1335 23.1215 14.1000 14.9509
100% 11.9787 17.1256 16.7657 23.1215 13.9000 14.6682

0% 22.4791 22.4791 26.9759 26.9759 21.2000 21.5154

25% 20.9223 22.4791 25.2845 26.9759 20.6500 21.2022
70% 50% 19.4692 22.4791 23.6853 26.9759 20.2692 20.8931
75% 18.1199 22.4791 22.1802 26.9759 20.1199 20.5883
100% 16.8728 22.4791 20.7699 26.9759 19.8728 20.2877

0% 27.8325 27.8325 30.8304 30.8304 26.8325 27.1901

25% 26.5043 27.8325 29.4595 30.8304 26.4900 26.9231
80% 50% 25.2176 27.8325 28.1256 30.8304 26.2176 26.6578
75% 23.9745 27.8325 26.8308 30.8304 26.1000 26.3942
100% 22.7769 27.8325 25.5769 30.8304 25.9000 26.1322
71
5.4 Summary

The four equations representing the ASME and ISO standards all resulted in differing

composite thickness recommendations for the same testing scenarios. Equation 1, ASME

PCC-2 Live Pressure Considered, violated the 0.25% max hoop strain requirement on

pipes with for corrosion percentages greater than 60%. Furthermore, as the live wrapping

pressure increased the strain rates increased as well. This means that the live pressure

considerations discounted the composite thickness too much. A perfect live pressure

thickness discount would result in equal hoop strains across all live wrapping pressures.

Equation 2, ASME PCC-2 No Live Pressure Considered, did not violate the 0.25%

max hoop strain requirement for any corrosion level or live wrapping pressure. While it

satisfied the main constraint of the code, it still does not fully serve its purpose of matching

0.25% hoop strain across all corrosion rates and wrapping pressures. The majority of

testing scenarios produce overly conservative thickness recommendations, especially at

lower corrosion percentages. Furthermore, the results from Equation 2 all experienced

reduction in strain rates as the live wrapping pressure increased. Since Equation 2

recommended the same composite thickness for each set of live pressures, and that the

hoop strain decreased as the live pressure increased, it can be deduced that live pressure

does play a role in determining the necessary composite thickness. ASME and PCC-2 need

a live pressure component of their analytical solutions.

Equation 3, ISO 24817 Live Pressure Considered, violated the max hoop strain

requirement for the most aggressive testing scenario (80% corroded wall, 100% live

wrapping pressure). For all other testing scenarios the ISO standard produced overly
72
conservative pipe wraps, considerably more so than Equation 2. Equation 3 suffers from

issues on both ends of the spectrum, as it is not conservative enough for edge cases like

80% wall thinning and 100% live wrapping pressure as well as being too conservative for

every other testing scenario.

Equation 33, ISO 24817 No Live Pressure Considered does not violate any max hoop

strain conditions on any wrapping pressure or corrosion percentage, but Equation 33 is the

most overly conservative composite thickness equation of the standards. Between 40%

and 70% wall thinning, the resulting hoop strain percentages hover around 0.16-0.17%

hoop strain. This is a considerable difference than the inputted 0.25% hoop strain these

equations are trying to match. The reason ISO values are so much more conservative than

the ASME standard is directly attributed to only allowing 72% yield stress to be modelled

in the steel yielding equation. This artificially forces the composites to be conservative for

the sake of being conservative. The ISO equations should not be viewed as attempts to

directly discern the resulting hoop strain.

Equation 32, the best fit parametric version of Equation 1, resulted in all of the testing

scenarios safely conforming to the max hoop strain limit. Furthermore the parametric

model produced the closest values to 0.25% hoop strain across all corrosion percentages

and live pressures. There is still not perfect agreeance between the parametric model a

consistent 0.25% hoop strains across every corrosion level and wrapping pressure. The

best fit parametric model had B values of 1.06 and C values of 0.20. These rates tend to

suggest that the ASME codes need to reexamine the standards to account for these

inconsistencies. It is theorized that the steel yielding term, Ys, should be increased to a

73
steel stress equivalent when strain hardened to no greater than 0.25% strain. This would

better reflect the 6% increase in the steel stress term the parametric model recommends.

As to why the C values need to be reduced to 20% of the current term, there is not as

obvious of an answer at this time. Given no straightforward answer, the author chooses

not to speculate but consider it a starting point for further research.

The overall best performing model was the optimized finite element model. Using

python scripting, it iteratively solved for the perfect composite thickness that created a

perfect 0.25% hoop strain. While this model performed in an outstanding fashion, it does

not serve the professional standards well. The standards need an analytical solution that

technicians and operators can easily use in the field that is clearly defined. Relying on

finite element model does not offer as concrete of a final answer as an analytical solution.

74
 CHAPTER VI

CONCLUSIONS AND FUTURE WORK

6.1 Conclusions

Based off the results and discussion the following conclusions are revealed. ASME

PCC-2 Substrate Yields Equation with Live Pressure is not conservative enough for safe

use, especially for significant wall thinning. It is not recommend this equation be used to

develop composite pipe wrap. ASME PCC-2 Substrate Yields Equation with No Live

Pressure Considerations allows for the creation of safe pipe wraps but is overly

conservative. The data also supports a live pressure component be included. It is not

recommend this equation be used to develop composite pipe wraps.

ISO 24817 Substrate Yields Equation with Live Pressure was also not conservative

enough for safe use. It is not recommend this equation be used to develop composite pipe

wraps. ISO 24817 Substrate Yields Equation with No Live Pressure Considerations

created safe pipe wraps but were significantly over conservative to the point of majorly

effecting the financial viability of a composite wrap. It is not recommend this equation be

used to develop composite pipe wraps.

The best performing model was the optimized finite element model which used python

scripts to solve for the perfect composite thickness. While this ensured no waste on code

compliant wraps, it is not feasible for professional standards to refer to a finite element

model over an analytical solution. It is not recommended that professional standards use

75
the finite element model to recommend and enforce composite wraps with live pressure

considerations.

The best fit parametric model performed better than all the standards solutions as it

created consistently safe wraps while recommending the least amount of excessive

wrapping. Furthermore, the best fit parametric model revealed that the steel yield strength

term within all the analytical solutions would be better served as a strain hardening stress

value greater than the yield stress. An in depth analysis of the theory governing load

transfer in composite wraps validated that parametric modelling is a necessity as the exact

composite thickness cannot be analytically solved for. Given these conclusions, it is the

final recommendation for the ASME PCC-2 and ISO 24817 to adopt the best fit parametric

model to create safe, resource efficient composite wraps.

6.2 Future Work

While this research has established new insights into how live pressure and wall

thinning effect composite repair hoop strains, it unveiled more questions along with its

answers. There is plenty of opportunity for follow up on this topic. Currently, it is not fully

understood why the parametric values selected produce the best fit. The parametric

formula as a whole should be vetted with real world pressure testing. Furthermore, these

models represent a level of abstraction from real world corrosion scenarios as the

corrosion is modelled as perfect thinning around the whole pipe. It is recommended to

continue testing the ISO, ASME and the best fit parametric model with localized machined

defects.

76
 REFERENCES

[1] Repair of Pressure Equipment and Piping ASME PCC-2-2015. New York, NY: The

American Society of Mechanical Engineers, 2015.

[2] “The role of pipeline age in pipeline safety - File.aspx.” [Online]. Available:

http://www.ingaa.org/File.aspx?id=19307. [Accessed: 13-Jun-2016].

[3] M. D. Chapetti, J. L. Otegui, C. Manfredi, and C. F. Martins, “Full scale experimental

analysis of stress states in sleeve repairs of gas pipelines,” International Journal of

Pressure Vessels and Piping, vol. 78, no. 5, pp. 379–387, May 2001.

[4] J. M. Duell, J. M. Wilson, and M. R. Kessler, “Analysis of a carbon composite

overwrap pipeline repair system,” International Journal of Pressure Vessels and

Piping, vol. 85, no. 11, pp. 782–788, Nov. 2008.

[5] J. l. f. Freire, R. d. Vieira, J. l. c. Diniz, and L. c. Meniconi, “Part 7: Effectiveness of

composite repairs applied to damaged pipeline,” Experimental Techniques, vol. 31,

no. 5, pp. 59–66, Sep. 2007.

[6] M. L. da Silva and H. da Costa Mattos, “Failure pressure estimations for corroded

pipelines,” Materials Science Forum, vol. 758, pp. 65–76, Jun. 2013.

[7] N. Saeed, H. Ronagh, and A. Virk, “Composite repair of pipelines, considering the

effect of live pressure-analytical and numerical models with respect to ISO/TS 24817

and ASME PCC-2,” Composites Part B, vol. 58, pp. 605–610, Mar. 2014.

77
[8] R. A. Esmaeel, M. A. Khan, and F. Taheri, “Assessment of the environmental effects

on the performance of FRP repaired steel pipes subjected to internal pressure,” J.

Pressure Vessel Technol, vol. 134, no. 4, pp. 041702–041702, Jul. 2012.

[9] W. K. Goertzen and M. R. Kessler, “Dynamic mechanical analysis of carbon/epoxy

composites for structural pipeline repair,” Composites Part B: Engineering, vol. 38,

no. 1, pp. 1–9, Jan. 2007.

[10] H. S. da Costa Mattos, J. M. L. Reis, L. M. Paim, M. L. da Silva, F. C. Amorim, and

V. A. Perrut, “Analysis of a glass fibre reinforced polyurethane composite repair

system for corroded pipelines at elevated temperatures,” Composite Structures, vol.

114, pp. 117–123, Aug. 2014.

[11] W. K. Goertzen and M. R. Kessler, “Creep behavior of carbon fiber/epoxy matrix

composites,” Materials Science and Engineering: A, vol. 421, no. 1–2, pp. 217–225,

Apr. 2006.

[12] M. W. Keller, B. D. Jellison, and T. Ellison, “Moisture effects on the thermal and

creep performance of carbon fiber/epoxy composites for structural pipeline repair,”

Composites Part B: Engineering, vol. 45, no. 1, pp. 1173–1180, Feb. 2013.

[13] M. Shamsuddoha, M. M. Islam, T. Aravinthan, A. Manalo, and K. Lau,

“Effectiveness of using fibre-reinforced polymer composites for underwater steel

pipeline repairs,” Composite Structures, vol. 100, pp. 40–54, Jun. 2013.

[14] C. Alexander and O. O. Ochoa, “Extending onshore pipeline repair to offshore steel

risers with carbon–fiber reinforced composites,” Composite Structures, vol. 92, no.

2, pp. 499–507, Jan. 2010.

78
[15] J. N. Reddy, Introduction to the finite element method, Third Edition. McGraw Hill,

pp. 100-149.

79
 APPENDIX I

EXPLANATION OF THIN WALL PRESSURE VESSELS

Given that a cylindrical pressure vessel has a radius to thickness ratio greater than 10,

the given assumptions of a thin walled pressure vessel apply. The thin wall assumption

states there is negligible transverse stress within the wall, thus the outside radius is equal

to the interior radius. Only hoop and axial stresses are considered for internally pressurized

vessels.

The axial stress is solved by analyzing the pressures exerted on the end cap of a

pressure vessel. Internal pressure P results in a longitudinal stress in the cylinder. The

force exerted on the pressure vessel endcap is simply the pressure multiplied by the area.

 * D2
F  P( ) (44)
4

The reactionary force experienced by the steel wall is equal to the average longitudinal

stress inside the wall multiplied by the cross sectional area of the pipe wall.

F   2 (2 rt ) (45)

Setting these forces equal to one another establishes an equation for the longitudinal stress.

PD
2  (46)
4t

Cutting the cylinder in the parallel to the longitudinal axis reveals the hoop stress

interactions. The internal pressure P results in a hoop stress. The force exerted upon the

half shell by the internal pressure is simply the pressure multiplied by the surface area.

80
 F  P( D * L) (47)

The reactionary force experienced by the steel is equal to the average hoop stress

multiplied by the area.

F   1 (2t * L) (48)

Setting these forces equal to one another establishes an equation for hoop stress.

PD
1  (49)
2t

These derivations form the basis for Barlow’s formula which relates the internal pressure

of a pipe to the strength of the material. Barlow’s formula is a reduced version of Lame’s

equations for pressure vessels.

81
 APPENDIX II

FINITE ELEMENT INPUTS FOR EACH TESTED STANDARDS EQUATION

Table 7. ABAQUS inputs for ASME PCC-2 live pressure considered (0.25% hoop strain)
ABAQUS inputs for ASME PCC-2 live pressure considered (0.25% hoop strain)
Live
Corrosion Steel Composite Live Design
Pressure
Percentage Thickness Thickness Pressure Pressure
Percentage
(%) (mm) (mm) (MPa) (MPa)
(%)
0% 4.977 1.0654 0 27.25
25% 4.977 0.9202 4.7688 27.25
30% 50% 4.977 0.8093 9.5375 27.25
75% 4.977 0.722 14.3063 27.25
100% 4.977 0.6516 19.075 27.25
0% 4.266 6.4188 0 27.25
25% 4.266 5.6337 4.0875 27.25
40% 50% 4.266 5.0029 8.175 27.25
75% 4.266 4.4889 12.2625 27.25
100% 4.266 4.0642 16.35 27.25
0% 3.555 11.7722 0 27.25
25% 3.555 10.5178 3.4067 27.25
50% 50% 3.555 9.4558 6.8134 27.25
75% 3.555 8.5549 10.2201 27.25
100% 3.555 7.7877 13.6269 27.25
0% 2.844 17.1256 0 27.25
25% 2.844 15.6034 2.75 27.25
60% 50% 2.844 14.2469 5.45 27.25
75% 2.844 13.0435 8.175 27.25
100% 2.844 11.9787 10.9 27.25

0% 2.133 22.4791 0 27.25

25% 2.133 20.9223 2.0440 27.25
70% 50% 2.133 19.4692 4.0881 27.25
75% 2.133 18.1199 6.1321 27.25
100% 2.133 16.8728 8.1761 27.25
0% 1.422 27.8325 0 27.25
25% 1.422 26.5043 1.3627 27.25
80% 50% 1.422 25.2176 2.7254 27.25
75% 1.422 23.9745 4.0881 27.25
100% 1.422 22.7769 5.4507 27.25
82
Table 8. ABAQUS inputs for ASME PCC-2 no live pressure considered (0.25% hoop strain)
ABAQUS inputs for ASME PCC-2 no live pressure considered (0.25% hoop strain)
Live
Corrosion Steel Composite Live Design
Pressure
Percentage Thickness Thickness Pressure Pressure
Percentage
(%) (mm) (mm) (MPa) (MPa)
(%)
0% 4.977 1.0654 0 27.25
25% 4.977 1.0654 4.7688 27.25
30% 50% 4.977 1.0654 9.5375 27.25
75% 4.977 1.0654 14.3063 27.25
100% 4.977 1.0654 19.0750 27.25

0% 4.266 6.4188 0 27.25

25% 4.266 6.4188 4.0875 27.25
40% 50% 4.266 6.4188 8.1750 27.25
75% 4.266 6.4188 12.2625 27.25
100% 4.266 6.4188 16.3500 27.25

0% 3.555 11.7722 0 27.25

25% 3.555 11.7722 3.4067 27.25
50% 50% 3.555 11.7722 6.8134 27.25
75% 3.555 11.7722 10.2201 27.25
100% 3.555 11.7722 13.6269 27.25

0% 2.844 17.1256 0 27.25

25% 2.844 17.1256 2.7500 27.25
60% 50% 2.844 17.1256 5.4500 27.25
75% 2.844 17.1256 8.1750 27.25
100% 2.844 17.1256 10.9000 27.25

0% 2.133 22.4791 0 27.25

25% 2.133 22.4791 2.0440 27.25
70% 50% 2.133 22.4791 4.0881 27.25
75% 2.133 22.4791 6.1321 27.25
100% 2.133 22.4791 8.1761 27.25

0% 1.422 27.8325 0 27.25

25% 1.422 27.8325 1.3627 27.25
80% 50% 1.422 27.8325 2.7254 27.25
75% 1.422 27.8325 4.0881 27.25
100% 1.422 27.8325 5.4507 27.25

83
Table 9. ABAQUS inputs for ISO 24817 live pressure considered (0.25% hoop strain)
ABAQUS inputs for ISO 24817 live pressure considered (0.25% hoop strain)
Live
Corrosion Steel Composite Live Design
Pressure
Percentage Thickness Thickness Pressure Pressure
Percentage
(%) (mm) (mm) (MPa) (MPa)
(%)
0% 4.977 11.5581 0.00001 27.25
25% 4.977 10.2324 4.7688 27.25
30% 50% 4.977 9.1391 9.5375 27.25
75% 4.977 8.231 14.3063 27.25
100% 4.977 7.4699 19.075 27.25

0% 4.266 15.4125 0.00001 27.25

25% 4.266 13.8053 4.0875 27.25
40% 50% 4.266 12.4349 8.175 27.25
75% 4.266 11.2657 12.2625 27.25
100% 4.266 10.2651 16.35 27.25

0% 3.555 19.267 0.00001 27.25

25% 3.555 17.4882 3.4063 27.25
50% 50% 3.555 15.9178 6.8125 27.25
75% 3.555 14.5365 10.2188 27.25
100% 3.555 13.3236 13.625 27.25

0% 2.844 23.1215 0.00001 27.25

25% 2.844 21.3051 2.725 27.25
60% 50% 2.844 19.6441 5.45 27.25
75% 2.844 18.1335 8.175 27.25
100% 2.844 16.7657 10.9 27.25

0% 2.133 26.9759 0.00001 27.25

25% 2.133 25.2845 2.0437 27.25
70% 50% 2.133 23.6853 4.0875 27.25
75% 2.133 22.1802 6.1312 27.25
100% 2.133 20.7699 8.175 27.25

0% 1.422 30.8304 0.00001 27.25

25% 1.422 29.4595 1.3625 27.25
80% 50% 1.422 28.1256 2.725 27.25
75% 1.422 26.8308 4.0875 27.25
100% 1.422 25.5769 5.45 27.25

84
Table 10. ABAQUS inputs for ISO 24817 no live pressure considered (0.25% hoop strain)
ABAQUS inputs for ISO 24817 no live pressure considered (0.25% hoop strain)
Live
Corrosion Steel Composite Live Design
Pressure
Percentage Thickness Thickness Pressure Pressure
Percentage
(%) (mm) (mm) (MPa) (MPa)
(%)
0% 4.977 1.0654 0 27.25
25% 4.977 1.0654 4.7688 27.25
30% 50% 4.977 1.0654 9.5375 27.25
75% 4.977 1.0654 14.3063 27.25
100% 4.977 1.0654 19.0750 27.25

0% 4.266 6.4188 0 27.25

25% 4.266 6.4188 4.0875 27.25
40% 50% 4.266 6.4188 8.1750 27.25
75% 4.266 6.4188 12.2625 27.25
100% 4.266 6.4188 16.3500 27.25

0% 3.555 11.7722 0 27.25

25% 3.555 11.7722 3.4067 27.25
50% 50% 3.555 11.7722 6.8134 27.25
75% 3.555 11.7722 10.2201 27.25
100% 3.555 11.7722 13.6269 27.25

0% 2.844 17.1256 0 27.25

25% 2.844 17.1256 2.7500 27.25
60% 50% 2.844 17.1256 5.4500 27.25
75% 2.844 17.1256 8.1750 27.25
100% 2.844 17.1256 10.9000 27.25

0% 2.133 22.4791 0 27.25

25% 2.133 22.4791 2.0440 27.25
70% 50% 2.133 22.4791 4.0881 27.25
75% 2.133 22.4791 6.1321 27.25
100% 2.133 22.4791 8.1761 27.25

0% 1.422 27.8325 0 27.25

25% 1.422 27.8325 1.3627 27.25
80% 50% 1.422 27.8325 2.7254 27.25
75% 1.422 27.8325 4.0881 27.25
100% 1.422 27.8325 5.4507 27.25

85